Get Your YouTube Summary:
- Copy the video link.
- Paste it into the input field.
- Click 'Summarize' to get your summary with timestamps.
Important Note on Subtitles:
- Automatic summary generation requires English subtitles on the video.
- If the video has no English subtitles, the automatic download of the transcript using the link will fail.
- Manual Alternative: You can still get a summary!
- Find the transcript on YouTube (usually below the video description when viewed on a desktop browser).
- Copy the entire transcript text manually. (Need help finding/copying? Watch the 'Demo Video' linked at the top right of this page).
- (Optional) Add any additional instructions after the transcript (e.g., 'Translate the summary to German.', 'Add a glossary of medical terms and jargon to the summary.').
For videos longer than 20 minutes:
- Select a Pro model for automatic summarization. Note that Pro usage is limited daily.
- If the Pro limit is reached (or if you prefer using your own tool), use the Copy Prompt button, paste the prompt into your AI tool, and run it there.
- 0:10 Xi Jinping's WWII History Claims Rebutted by Taiwan: Xi Jinping made statements about Taiwan's return to China being a WWII outcome based on the Cairo and Potsdam declarations. Taiwan's Mainland Affairs Council (MAC) refuted this, stating the PRC did not exist at the time of these declarations and that Taiwan was handed to the Republic of China (ROC), not the PRC. UN Resolution 2758 also did not address Taiwan's sovereignty.
- 3:03 Lai Ching-te Commemorates V-E Day: Taiwanese President Lai Ching-te held Taiwan's first government-level V-E Day commemoration, emphasizing unity against aggression, the prosperity of democracies, and the need for the free world to stand against threats. The commentator praises this as a positive step for Taiwan asserting its sovereignty.
- 11:43 KMT's Chu Li-luan Likens Lai to Hitler: KMT Chairman Chu Li-luan compared Lai Ching-te and the DPP to Hitler and Nazis. This drew criticism from several European representatives. The commentator views this as a deliberate, CCP-aligned tactic to provide a pretext for potential Chinese aggression, similar to Putin's "denazification" justification in Ukraine.
- 18:44 India-Pakistan Tensions and Conflicting Claims: Both India and Pakistan make claims and counter-claims about shooting down each other's fighter jets. U.S. officials reportedly confirmed Pakistan used Chinese J-10s to down at least two Indian jets. The commentator believes Pakistan likely downed three Indian jets, and India is matching this claim for a show of parity.
- 29:11 Ukraine War: Russian Victory Day Preparations Disrupted: A Russian airbase near Moscow, home to aerobatic teams, was reportedly hit by Ukrainian drones. Russia implemented unprecedented, large-scale internet shutdowns across 60% of its territory, citing security for Victory Day, causing significant disruption.
- 32:52 Trump Calls for Ukraine Ceasefire, EU Support for Kyiv: Trump posted calling for a 30-day unconditional ceasefire in Ukraine. Meanwhile, EU foreign ministers are visiting Kyiv, announcing €1 billion in aid and a new sanctions package against Russia.
- 35:01 New American Pope Elected (Leo XIV): Robert Prevost from Chicago was elected as the new Pope. While some in the U.S. (like Vance) welcomed an American Pope, MAGA figures like Laura Loomer criticized him as anti-Trump and a "Marxist puppet."
- 40:18 Germany Reflects on V-E Day, Condemns Russian Aggression: German President Steinmeier thanked the Allies for defeating Nazism and praised the Soviet Red Army's historical role, while strongly condemning Russia's current war in Ukraine and its false "denazification" narrative.
- 53:35 Trump Signs "Framework Announcement" with UK on Trade: Trump and the UK PM announced a trade deal, with the US lowering tariffs on UK cars (25% to 10%) and steel/aluminum (to 0% with quotas), while the UK opens markets to some US goods. Bloomberg's analysis, cited by the commentator, suggests it's more of a symbolic "framework announcement" than a substantive deal, primarily a political win for Trump.
- 1:04:50 US Automakers Criticize US-UK Deal: Major US auto manufacturers like GM and Ford criticized the deal, arguing it harms US interests by favoring UK imports over those from North American partners under USMCA.
- 1:13:30 Xi-Putin Sign "Big Deal" Lacking Substance: On the same day as the US-UK announcement, Xi and Putin signed over 20 agreements. However, the commentator notes a lack of concrete details, especially regarding energy (e.g., the Siberian gas pipeline). It's suggested Xi is keeping his options open amid nascent talks with the U.S.
- 1:20:13 WSJ Report: US-China Re-engagement via Fentanyl Talks: According to the Wall Street Journal, US-China dialogue resumed with China inquiring about U.S. expectations on fentanyl control, a reversal of Xi's earlier misjudgment that fentanyl was merely a U.S. pretext. This reportedly led to Lighthizer and He Lifeng agreeing to meet in Switzerland.
- 1:26:08 NY Post: Trump May Cut China Tariffs to 50-54%: The New York Post reported that the Trump administration is considering cutting the 145% tariffs on Chinese goods to 50-54% as early as next week if weekend talks in Switzerland go well. This would be an interim measure while a full trade deal is negotiated.
- 1:31:52 Commentator's Outlook on US-China Trade Talks: The weekend meeting is likely for ice-breaking and setting principles. A potential outcome is the US reducing tariffs to around 50-54% while China removes retaliatory tariffs, followed by protracted negotiations. The commentator believes China has "no choice" but to accept harsh terms due to its economic situation.
- 00:00:02 Introduction to LLM Benchmarks & Evaluation Challenges: The video begins by noting the increasing benchmark scores of new LLM releases (e.g., MMLU, GPQA for multiple choice; HumanEval for code). It highlights the difficulty in evaluating subjective tasks, such as explaining a concept clearly to a specific audience, and sets out to provide an overview of evaluation methods for such tasks.
- 00:00:52 Sponsor Segment: VoiceWriter.io: Introduces VoiceWriter, an AI-powered tool designed to help users write faster using voice. It features real-time grammar correction, punctuation addition, advanced speech recognition, and a Chrome extension. A free trial is offered.
- 00:01:22 Category 1: Reference-Based Evaluation:
- This method is useful when a reference (ideal output) is available, common in tasks like summarization and machine translation. It's best for well-defined answers with limited variability.
- 00:01:52 BLEU and ROUGE: These are traditional metrics that calculate the overlap of n-grams between the model's output and the reference. They perform surface-level matching and are not very nuanced, treating tokens as simply correct or incorrect without judging the degree of error.
- 00:02:39 Advanced Model-Based Reference Scoring: More sophisticated approaches use language models.
- 00:02:45 BERTScore: Employs an encoder LM to get embeddings for words in both candidate and reference sentences, then scores the similarity of these embedding sequences.
- 00:03:09 COMET and BLURT: These metrics involve training a model to predict how a human would rate the similarity between two sentences. COMET, for example, encodes both sentences and uses a regression model for this prediction.
- 00:03:26 Category 2: Human Evaluation:
- Involves humans directly assessing LLM outputs. Methods include rating output quality or, increasingly popular, pairwise comparison where users select the preferred of two model responses.
- 00:03:54 LMSys Chatbot Arena: An open-source platform where users query two random models and pick the better response. Aggregated ratings generate an ELO score, creating a model leaderboard.
- 00:04:42 Pros and Cons: While providing valuable insight, human evaluation is expensive and annotators aren't perfect, often influenced by superficial aspects like output length, formatting (e.g., use of bold text, emojis).
- 00:05:31 Solution - Style Control: To address biases, a regression model can be fitted to account for style attributes (length, formatting) that might influence user preference, thereby disentangling style from true content quality. This showed some models (e.g., Claude 3.5) perform better after style control, while others (e.g., GPT-4o mini) perform worse, indicating their wins were sometimes due to formatting.
- 00:06:34 Category 3: LLM as a Judge:
- This approach uses another LLM to evaluate the output(s) of one or more LLMs.
- 00:06:57 Setup: The judge LLM can pick the better of two outputs or rate outputs on predefined criteria (e.g., helpfulness, relevance, accuracy, depth, creativity). Multi-factor rating provides deeper insights. (Example given: VoiceWriter uses LLMs to evaluate style adherence and content preservation for its grammar correction tool).
- 00:08:07 Validation: The MT-Bench paper found that GPT-4's agreement with human judgments (85%) was even higher than human-human agreement (81%), suggesting LLMs can be reliable judges.
- 00:08:39 Potential Biases: LLM judges can suffer from biases similar to humans, such as length bias (favoring longer responses), position bias (favoring the first response shown), or self-preference bias (a model rating its own outputs higher). These need to be controlled.
- 00:09:09 Cost and Utility: While using LLMs for evaluation has a cost, it's typically cheaper than human evaluation, with a range of models offering different cost-quality trade-offs. LLMs are also useful for evaluating structured short answers where rule-based evaluation is hard (e.g., OmniMath benchmark for mathematical expressions).
- 00:09:53 Key Takeaway: Evaluating LLMs on subjective tasks requires moving beyond simple metrics to a combination of reference-based, human, and LLM-as-judge evaluations. Each method has its strengths, weaknesses, and associated costs, and understanding biases (and how to control for them) is crucial for accurate assessment.
- 0:31 Introduction to Finale's End: The video focuses on Finale, a music notation software discontinued in August 2024, as a case study on how creative software dies, specifically due to failure to address long-standing usability issues and evolving market dynamics.
- 6:49 Pioneering Origins: Finale was created by Coda Music in 1988 by Phil Farrand and John Borowicz, causing a sensation at NAMM, particularly for its real-time MIDI keyboard input feature.
- 7:43 Early Success and Innovation: Despite a high price (around £3,170 adjusted), Finale became a fully-rounded notation app, defining core features still used today (compose, playback, professional publishing) and supporting the new MIDI protocol.
- 9:03 Early User Interface: The initial UI was described as having a "clean simplicity" inspired by contemporary Mac apps, including tools like the "mass mover" for copying phrases.
- 9:50 Initial Difficulty: From the start, Finale was acknowledged as not being the most straightforward application, often requiring users to spend significant time learning it, utilizing dialog boxes and tutorials.
- 12:00 Rivalry with Sibelius: Finale competed directly with Sibelius (emerging around 1990) and Igor Engraver. Sibelius gained a reputation for being more intuitive, eventually surpassing Finale in ease of use.
- 13:04 Declining Development Pace: Over the 2000s, Finale's development slowed, and its interface felt increasingly outdated compared to modern software.
- 13:40 User Experience Critique (Version 27): The video critiques the modern version's interface as "malnourished," with a "jaundiced colour" and "joyless grey background," arguing its dated appearance contributed to its decline.
- 17:23 Frustrating Font Changes: Changing fonts is highlighted as an incredibly difficult and frustrating process, requiring navigating nested dialogs, holding modifier keys, and encountering limitations like changes only applying to future items.
- 19:36 Instrument Name Font Issues: Changing fonts for instrument names is particularly complex, requiring finding hidden anchor boxes, navigating multiple dialogs, and in some cases, deleting and re-adding the instrument entirely due to missing options (21:08).
- 22:09 Difficult Management Dialogs: Accessing the ScoreManager for adding/deleting instruments involves a dense, unintuitive dialog, and even basic tasks like moving instruments require discovering non-obvious drag handles (23:04).
- 24:12 Tempo Marking Complexity: Editing tempo markings also involves navigating confusing dialogs and understanding specialist jargon (25:04).
- 26:23 Power User Bias Critique: The narrator argues that creative software, including notation apps, often caters too much to power users who rely heavily on shortcuts, neglecting novice and intermediate users who primarily use the UI (28:36). Research is cited showing most users only use a few shortcuts (28:48).
- 32:02 Painful Note Entry (Simple Entry): Finale's "Simple Entry" mode is criticized as highly inflexible and unintuitive, preventing placing notes freely, requiring mode switching to see rests, and behaving like text editing where deleting notes shifts subsequent notes backward (35:06).
- 39:04 Speedy Entry Mode: "Speedy Entry" mode is found to be more straightforward, especially with a MIDI keyboard, though it can still result in notes "bunching up" visually until exiting the mode (40:31). A positive feature is the ability to drag-copy notes/sections (40:54).
- 41:11 Complex Markings: Adding standard musical markings (pedal lines, string numbers, dynamics, etc.) involves convoluted, non-intuitive workflows hidden behind obscure menus, modifier keys, and layered dialogs.
- 48:26 Dynamics/Hairpin Difficulty: Drawing hairpins is demonstrated as a particularly unintuitive process, requiring a rare "double-click and hold and drag" interaction pattern because normal click-and-drag is reserved for selection (51:06).
- 52:05 User Testing: Brief user testing with other YouTubers confirms that even seasoned musicians find basic tasks in Finale difficult and non-obvious.
- 53:56 Layout and Spacing Challenges: Finale requires significant manual intervention for layout and spacing, although it offers precise tools (Measure tool, segment width) once the user understands complex shortcuts and workflows (55:27).
- 57:11 Finale's Paradox: While brutally difficult to learn, mastering Finale provides flexibility, making it arguably the most flexible but also the most time-consuming of major notation apps.
- 57:44 Reasons for Decline (Organizational & Technical): The video explores the behind-the-scenes factors: an ancient codebase that was hard to update (58:25), dependence on third-party plugins that broke with updates (59:36), strategic shifts at parent company MakeMusic/Peaksware prioritizing the educational SmartMusic platform (1:01:22).
- 1:06:39 Shift to MakeMusicCloud: The acquisition of Weezic (browser-based, MusicXML) and Alfred (publisher) led to the creation of MakeMusicCloud, which used MusicXML, eliminating Finale's role as a necessary core technology (1:07:19).
- 1:07:51 MuseScore's Growth: MakeMusic's decision to charge for the beginner version Finale Notepad boosted downloads of free competitor MuseScore, steering younger users away from the Finale ecosystem.
- 1:08:38 Discontinuation Announcement and Backlash: The August 2024 announcement initially included shutting down the software's verification system, causing outrage among users and publishers with Finale catalogues, leading to a quick reversal to allow indefinite authorization (1:09:07).
- 1:18:12 Common Bugs and Quirks: The video concludes with a montage of common, frustrating bugs and inconsistencies experienced by Finale users, highlighting the unrepaired state of the software.
identifier: 2562
model: gemini-2.5-pro-exp-03-25| input-price: 1.25 output-price: 10.0 max-context-length: 128_000
host: 193.8.40.126
https://www.youtube.com/watch?v=WVG4C0UwQ-sinclude_comments: None
include_timestamps: 1
include_glossary: None
output_language: en
cost: 0.06020125
Okay, here is the abstract and summary for the provided transcript: *Abstract:* This live stream commentary by "大康有话说" on May 9, 2025, covers several geopolitical developments. Key topics include Xi Jinping's controversial remarks on World War II history, which were promptly rebutted by Taiwan, and Taiwan's Lai Ching-te holding the nation's first official V-E Day commemoration. The commentator analyzes the KMT's comparison of Lai to Hitler as a CCP-aligned tactic. The stream also discusses President Trump's signing of a "framework announcement" with the UK on trade, which the commentator, citing Bloomberg, views as largely symbolic. This is contrasted with a Xi-Putin meeting on the same day, which reportedly yielded a "big deal" that the commentator finds lacking in substance, suggesting Xi is keeping options open with the U.S. The report also delves into renewed U.S.-China engagement, supposedly initiated via discussions on fentanyl control, leading to a potential meeting in Switzerland and a New York Post report that Trump might reduce tariffs on Chinese goods to 50-54% soon. Other discussed events include ongoing India-Pakistan tensions with conflicting claims of downed aircraft, developments in the Ukraine war including a drone attack on a Russian airbase and large-scale internet shutdowns in Russia, the election of a new American Pope, and Germany's V-E Day reflections. *Key Developments and Commentary from the Livestream* * *0:10 Xi Jinping's WWII History Claims Rebutted by Taiwan:* Xi Jinping made statements about Taiwan's return to China being a WWII outcome based on the Cairo and Potsdam declarations. Taiwan's Mainland Affairs Council (MAC) refuted this, stating the PRC did not exist at the time of these declarations and that Taiwan was handed to the Republic of China (ROC), not the PRC. UN Resolution 2758 also did not address Taiwan's sovereignty. * *3:03 Lai Ching-te Commemorates V-E Day:* Taiwanese President Lai Ching-te held Taiwan's first government-level V-E Day commemoration, emphasizing unity against aggression, the prosperity of democracies, and the need for the free world to stand against threats. The commentator praises this as a positive step for Taiwan asserting its sovereignty. * *11:43 KMT's Chu Li-luan Likens Lai to Hitler:* KMT Chairman Chu Li-luan compared Lai Ching-te and the DPP to Hitler and Nazis. This drew criticism from several European representatives. The commentator views this as a deliberate, CCP-aligned tactic to provide a pretext for potential Chinese aggression, similar to Putin's "denazification" justification in Ukraine. * *18:44 India-Pakistan Tensions and Conflicting Claims:* Both India and Pakistan make claims and counter-claims about shooting down each other's fighter jets. U.S. officials reportedly confirmed Pakistan used Chinese J-10s to down at least two Indian jets. The commentator believes Pakistan likely downed three Indian jets, and India is matching this claim for a show of parity. * *29:11 Ukraine War: Russian Victory Day Preparations Disrupted:* A Russian airbase near Moscow, home to aerobatic teams, was reportedly hit by Ukrainian drones. Russia implemented unprecedented, large-scale internet shutdowns across 60% of its territory, citing security for Victory Day, causing significant disruption. * *32:52 Trump Calls for Ukraine Ceasefire, EU Support for Kyiv:* Trump posted calling for a 30-day unconditional ceasefire in Ukraine. Meanwhile, EU foreign ministers are visiting Kyiv, announcing €1 billion in aid and a new sanctions package against Russia. * *35:01 New American Pope Elected (Leo XIV):* Robert Prevost from Chicago was elected as the new Pope. While some in the U.S. (like Vance) welcomed an American Pope, MAGA figures like Laura Loomer criticized him as anti-Trump and a "Marxist puppet." * *40:18 Germany Reflects on V-E Day, Condemns Russian Aggression:* German President Steinmeier thanked the Allies for defeating Nazism and praised the Soviet Red Army's historical role, while strongly condemning Russia's current war in Ukraine and its false "denazification" narrative. * *53:35 Trump Signs "Framework Announcement" with UK on Trade:* Trump and the UK PM announced a trade deal, with the US lowering tariffs on UK cars (25% to 10%) and steel/aluminum (to 0% with quotas), while the UK opens markets to some US goods. Bloomberg's analysis, cited by the commentator, suggests it's more of a symbolic "framework announcement" than a substantive deal, primarily a political win for Trump. * *1:04:50 US Automakers Criticize US-UK Deal:* Major US auto manufacturers like GM and Ford criticized the deal, arguing it harms US interests by favoring UK imports over those from North American partners under USMCA. * *1:13:30 Xi-Putin Sign "Big Deal" Lacking Substance:* On the same day as the US-UK announcement, Xi and Putin signed over 20 agreements. However, the commentator notes a lack of concrete details, especially regarding energy (e.g., the Siberian gas pipeline). It's suggested Xi is keeping his options open amid nascent talks with the U.S. * *1:20:13 WSJ Report: US-China Re-engagement via Fentanyl Talks:* According to the Wall Street Journal, US-China dialogue resumed with China inquiring about U.S. expectations on fentanyl control, a reversal of Xi's earlier misjudgment that fentanyl was merely a U.S. pretext. This reportedly led to Lighthizer and He Lifeng agreeing to meet in Switzerland. * *1:26:08 NY Post: Trump May Cut China Tariffs to 50-54%:* The New York Post reported that the Trump administration is considering cutting the 145% tariffs on Chinese goods to 50-54% as early as next week if weekend talks in Switzerland go well. This would be an interim measure while a full trade deal is negotiated. * *1:31:52 Commentator's Outlook on US-China Trade Talks:* The weekend meeting is likely for ice-breaking and setting principles. A potential outcome is the US reducing tariffs to around 50-54% while China removes retaliatory tariffs, followed by protracted negotiations. The commentator believes China has "no choice" but to accept harsh terms due to its economic situation. I used gemini-2.5-pro-exp-03-25| input-price: 1.25 output-price: 10.0 max-context-length: 128_000 on rocketrecap dot com to summarize the transcript. Cost (if I didn't use the free tier): $0.06 Input tokens: 36793 Output tokens: 1421
Okay, here is the abstract and summary for the provided transcript:
Abstract:
This live stream commentary by "大康有话说" on May 9, 2025, covers several geopolitical developments. Key topics include Xi Jinping's controversial remarks on World War II history, which were promptly rebutted by Taiwan, and Taiwan's Lai Ching-te holding the nation's first official V-E Day commemoration. The commentator analyzes the KMT's comparison of Lai to Hitler as a CCP-aligned tactic. The stream also discusses President Trump's signing of a "framework announcement" with the UK on trade, which the commentator, citing Bloomberg, views as largely symbolic. This is contrasted with a Xi-Putin meeting on the same day, which reportedly yielded a "big deal" that the commentator finds lacking in substance, suggesting Xi is keeping options open with the U.S. The report also delves into renewed U.S.-China engagement, supposedly initiated via discussions on fentanyl control, leading to a potential meeting in Switzerland and a New York Post report that Trump might reduce tariffs on Chinese goods to 50-54% soon. Other discussed events include ongoing India-Pakistan tensions with conflicting claims of downed aircraft, developments in the Ukraine war including a drone attack on a Russian airbase and large-scale internet shutdowns in Russia, the election of a new American Pope, and Germany's V-E Day reflections.
Key Developments and Commentary from the Livestream
Below, I will provide input for an example video (comprising of title, description, and transcript, in this order) and the corresponding abstract and summary I expect. Afterward, I will provide a new transcript that I want you to summarize in the same format.
**Please give an abstract of the transcript and then summarize the transcript in a self-contained bullet list format.** Include starting timestamps, important details and key takeaways.
Example Input:
Fluidigm Polaris Part 2- illuminator and camera
mikeselectricstuff
131K subscribers
Subscribed
369
Share
Download
Clip
Save
5,857 views Aug 26, 2024
Fluidigm Polaris part 1 : • Fluidigm Polaris (Part 1) - Biotech g...
Ebay listings: https://www.ebay.co.uk/usr/mikeselect...
Merch https://mikeselectricstuff.creator-sp...
Transcript
Follow along using the transcript.
Show transcript
mikeselectricstuff
131K subscribers
Videos
About
Support on Patreon
40 Comments
@robertwatsonbath
6 hours ago
Thanks Mike. Ooof! - with the level of bodgery going on around 15:48 I think shame would have made me do a board re spin, out of my own pocket if I had to.
1
Reply
@Muonium1
9 hours ago
The green LED looks different from the others and uses phosphor conversion because of the "green gap" problem where green InGaN emitters suffer efficiency droop at high currents. Phosphide based emitters don't start becoming efficient until around 600nm so also can't be used for high power green emitters. See the paper and plot by Matthias Auf der Maur in his 2015 paper on alloy fluctuations in InGaN as the cause of reduced external quantum efficiency at longer (green) wavelengths.
4
Reply
1 reply
@tafsirnahian669
10 hours ago (edited)
Can this be used as an astrophotography camera?
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
6 hours ago
Yes, but may need a shutter to avoid light during readout
Reply
@2010craggy
11 hours ago
Narrowband filters we use in Astronomy (Astrophotography) are sided- they work best passing light in one direction so I guess the arrows on the filter frames indicate which way round to install them in the filter wheel.
1
Reply
@vitukz
12 hours ago
A mate with Channel @extractions&ire could use it
2
Reply
@RobertGallop
19 hours ago
That LED module says it can go up to 28 amps!!! 21 amps for 100%. You should see what it does at 20 amps!
Reply
@Prophes0r
19 hours ago
I had an "Oh SHIT!" moment when I realized that the weird trapezoidal shape of that light guide was for keystone correction of the light source.
Very clever.
6
Reply
@OneBiOzZ
20 hours ago
given the cost of the CCD you think they could have run another PCB for it
9
Reply
@tekvax01
21 hours ago
$20 thousand dollars per minute of run time!
1
Reply
@tekvax01
22 hours ago
"We spared no expense!" John Hammond Jurassic Park.
*(that's why this thing costs the same as a 50-seat Greyhound Bus coach!)
Reply
@florianf4257
22 hours ago
The smearing on the image could be due to the fact that you don't use a shutter, so you see brighter stripes under bright areas of the image as you still iluminate these pixels while the sensor data ist shifted out towards the top. I experienced this effect back at university with a LN-Cooled CCD for Spectroscopy. The stripes disapeared as soon as you used the shutter instead of disabling it in the open position (but fokussing at 100ms integration time and continuous readout with a focal plane shutter isn't much fun).
12
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
12 hours ago
I didn't think of that, but makes sense
2
Reply
@douro20
22 hours ago (edited)
The red LED reminds me of one from Roithner Lasertechnik. I have a Symbol 2D scanner which uses two very bright LEDs from that company, one red and one red-orange. The red-orange is behind a lens which focuses it into an extremely narrow beam.
1
Reply
@RicoElectrico
23 hours ago
PFG is Pulse Flush Gate according to the datasheet.
Reply
@dcallan812
23 hours ago
Very interesting. 2x
Reply
@littleboot_
1 day ago
Cool interesting device
Reply
@dav1dbone
1 day ago
I've stripped large projectors, looks similar, wonder if some of those castings are a magnesium alloy?
Reply
@kevywevvy8833
1 day ago
ironic that some of those Phlatlight modules are used in some of the cheapest disco lights.
1
Reply
1 reply
@bill6255
1 day ago
Great vid - gets right into subject in title, its packed with information, wraps up quickly. Should get a YT award! imho
3
Reply
@JAKOB1977
1 day ago (edited)
The whole sensor module incl. a 5 grand 50mpix sensor for 49 £.. highest bid atm
Though also a limited CCD sensor, but for the right buyer its a steal at these relative low sums.
Architecture Full Frame CCD (Square Pixels)
Total Number of Pixels 8304 (H) × 6220 (V) = 51.6 Mp
Number of Effective Pixels 8208 (H) × 6164 (V) = 50.5 Mp
Number of Active Pixels 8176 (H) × 6132 (V) = 50.1 Mp
Pixel Size 6.0 m (H) × 6.0 m (V)
Active Image Size 49.1 mm (H) × 36.8 mm (V)
61.3 mm (Diagonal),
645 1.1x Optical Format
Aspect Ratio 4:3
Horizontal Outputs 4
Saturation Signal 40.3 ke−
Output Sensitivity 31 V/e−
Quantum Efficiency
KAF−50100−CAA
KAF−50100−AAA
KAF−50100−ABA (with Lens)
22%, 22%, 16% (Peak R, G, B)
25%
62%
Read Noise (f = 18 MHz) 12.5 e−
Dark Signal (T = 60°C) 42 pA/cm2
Dark Current Doubling Temperature 5.7°C
Dynamic Range (f = 18 MHz) 70.2 dB
Estimated Linear Dynamic Range
(f = 18 MHz)
69.3 dB
Charge Transfer Efficiency
Horizontal
Vertical
0.999995
0.999999
Blooming Protection
(4 ms Exposure Time)
800X Saturation Exposure
Maximum Date Rate 18 MHz
Package Ceramic PGA
Cover Glass MAR Coated, 2 Sides or
Clear Glass
Features
• TRUESENSE Transparent Gate Electrode
for High Sensitivity
• Ultra-High Resolution
• Board Dynamic Range
• Low Noise Architecture
• Large Active Imaging Area
Applications
• Digitization
• Mapping/Aerial
• Photography
• Scientific
Thx for the tear down Mike, always a joy
Reply
@martinalooksatthings
1 day ago
15:49 that is some great bodging on of caps, they really didn't want to respin that PCB huh
8
Reply
@RhythmGamer
1 day ago
Was depressed today and then a new mike video dropped and now I’m genuinely happy to get my tear down fix
1
Reply
@dine9093
1 day ago (edited)
Did you transfrom into Mr Blobby for a moment there?
2
Reply
@NickNorton
1 day ago
Thanks Mike. Your videos are always interesting.
5
Reply
@KeritechElectronics
1 day ago
Heavy optics indeed... Spare no expense, cost no object. Splendid build quality. The CCD is a thing of beauty!
1
Reply
@YSoreil
1 day ago
The pricing on that sensor is about right, I looked in to these many years ago when they were still in production since it's the only large sensor you could actually buy. Really cool to see one in the wild.
2
Reply
@snik2pl
1 day ago
That leds look like from led projector
Reply
@vincei4252
1 day ago
TDI = Time Domain Integration ?
1
Reply
@wolpumba4099
1 day ago (edited)
Maybe the camera should not be illuminated during readout.
From the datasheet of the sensor (Onsemi): saturation 40300 electrons, read noise 12.5 electrons per pixel @ 18MHz (quite bad). quantum efficiency 62% (if it has micro lenses), frame rate 1 Hz. lateral overflow drain to prevent blooming protects against 800x (factor increases linearly with exposure time) saturation exposure (32e6 electrons per pixel at 4ms exposure time), microlens has +/- 20 degree acceptance angle
i guess it would be good for astrophotography
4
Reply
@txm100
1 day ago (edited)
Babe wake up a new mikeselectricstuff has dropped!
9
Reply
@vincei4252
1 day ago
That looks like a finger-lakes filter wheel, however, for astronomy they'd never use such a large stepper.
1
Reply
@MRooodddvvv
1 day ago
yaaaaay ! more overcomplicated optical stuff !
4
Reply
1 reply
@NoPegs
1 day ago
He lives!
11
Reply
1 reply
Transcript
0:00
so I've stripped all the bits of the
0:01
optical system so basically we've got
0:03
the uh the camera
0:05
itself which is mounted on this uh very
0:09
complex
0:10
adjustment thing which obviously to set
0:13
you the various tilt and uh alignment
0:15
stuff then there's two of these massive
0:18
lenses I've taken one of these apart I
0:20
think there's something like about eight
0:22
or nine Optical elements in here these
0:25
don't seem to do a great deal in terms
0:26
of electr magnification they're obiously
0:28
just about getting the image to where it
0:29
uh where it needs to be just so that
0:33
goes like that then this Optical block I
0:36
originally thought this was made of some
0:37
s crazy heavy material but it's just
0:39
really the sum of all these Optical bits
0:41
are just ridiculously heavy those lenses
0:43
are about 4 kilos each and then there's
0:45
this very heavy very solid um piece that
0:47
goes in the middle and this is so this
0:49
is the filter wheel assembly with a
0:51
hilariously oversized steper
0:53
motor driving this wheel with these very
0:57
large narrow band filters so we've got
1:00
various different shades of uh
1:03
filters there five Al together that
1:06
one's actually just showing up a silver
1:07
that's actually a a red but fairly low
1:10
transmission orangey red blue green
1:15
there's an excess cover on this side so
1:16
the filters can be accessed and changed
1:19
without taking anything else apart even
1:21
this is like ridiculous it's like solid
1:23
aluminium this is just basically a cover
1:25
the actual wavelengths of these are um
1:27
488 525 570 630 and 700 NM not sure what
1:32
the suffix on that perhaps that's the uh
1:34
the width of the spectral line say these
1:37
are very narrow band filters most of
1:39
them are you very little light through
1:41
so it's still very tight narrow band to
1:43
match the um fluoresence of the dies
1:45
they're using in the biochemical process
1:48
and obviously to reject the light that's
1:49
being fired at it from that Illuminator
1:51
box and then there's a there's a second
1:53
one of these lenses then the actual sort
1:55
of samples below that so uh very serious
1:58
amount of very uh chunky heavy Optics
2:01
okay let's take a look at this light
2:02
source made by company Lumen Dynamics
2:04
who are now part of
2:06
excelitas self-contained unit power
2:08
connector USB and this which one of the
2:11
Cable Bundle said was a TTL interface
2:14
USB wasn't used in uh the fluid
2:17
application output here and I think this
2:19
is an input for um light feedback I
2:21
don't if it's regulated or just a measur
2:23
measurement facility and the uh fiber
2:27
assembly
2:29
Square Inlet there and then there's two
2:32
outputs which have uh lens assemblies
2:35
and this small one which goes back into
2:37
that small Port just Loops out of here
2:40
straight back in So on this side we've
2:42
got the electronics which look pretty
2:44
straightforward we've got a bit of power
2:45
supply stuff over here and we've got
2:48
separate drivers for each wavelength now
2:50
interesting this is clearly been very
2:52
specifically made for this application
2:54
you I was half expecting like say some
2:56
generic drivers that could be used for a
2:58
number of different things but actually
3:00
literally specified the exact wavelength
3:02
on the PCB there is provision here for
3:04
385 NM which isn't populated but this is
3:07
clearly been designed very specifically
3:09
so these four drivers look the same but
3:10
then there's two higher power ones for
3:12
575 and
3:14
520 a slightly bigger heat sink on this
3:16
575 section there a p 24 which is
3:20
providing USB interface USB isolator the
3:23
USB interface just presents as a comport
3:26
I did have a quick look but I didn't
3:27
actually get anything sensible um I did
3:29
dump the Pi code out and there's a few
3:31
you a few sort of commands that you
3:32
could see in text but I didn't actually
3:34
manage to get it working properly I
3:36
found some software for related version
3:38
but it didn't seem to want to talk to it
3:39
but um I say that wasn't used for the
3:41
original application it might be quite
3:42
interesting to get try and get the Run
3:44
hours count out of it and the TTL
3:46
interface looks fairly straightforward
3:48
we've got positions for six opto
3:50
isolators but only five five are
3:52
installed so that corresponds with the
3:54
unused thing so I think this hopefully
3:56
should be as simple as just providing a
3:57
ttrl signal for each color to uh enable
4:00
it a big heat sink here which is there I
4:03
think there's like a big S of metal
4:04
plate through the middle of this that
4:05
all the leads are mounted on the other
4:07
side so this is heat sinking it with a
4:09
air flow from a uh just a fan in here
4:13
obviously don't have the air flow
4:14
anywhere near the Optics so conduction
4:17
cool through to this plate that's then
4:18
uh air cooled got some pots which are
4:21
presumably power
4:22
adjustments okay let's take a look at
4:24
the other side which is uh much more
4:27
interesting see we've got some uh very
4:31
uh neatly Twisted cable assemblies there
4:35
a bunch of leads so we've got one here
4:37
475 up here 430 NM 630 575 and 520
4:44
filters and dcro mirrors a quick way to
4:48
see what's white is if we just shine
4:49
some white light through
4:51
here not sure how it is is to see on the
4:54
camera but shining white light we do
4:55
actually get a bit of red a bit of blue
4:57
some yellow here so the obstacle path
5:00
575 it goes sort of here bounces off
5:03
this mirror and goes out the 520 goes
5:07
sort of down here across here and up
5:09
there 630 goes basically straight
5:13
through
5:15
430 goes across there down there along
5:17
there and the 475 goes down here and
5:20
left this is the light sensing thing
5:22
think here there's just a um I think
5:24
there a photo diode or other sensor
5:26
haven't actually taken that off and
5:28
everything's fixed down to this chunk of
5:31
aluminium which acts as the heat
5:32
spreader that then conducts the heat to
5:33
the back side for the heat
5:35
sink and the actual lead packages all
5:38
look fairly similar except for this one
5:41
on the 575 which looks quite a bit more
5:44
substantial big spay
5:46
Terminals and the interface for this
5:48
turned out to be extremely simple it's
5:50
literally a 5V TTL level to enable each
5:54
color doesn't seem to be any tensity
5:56
control but there are some additional
5:58
pins on that connector that weren't used
5:59
in the through time thing so maybe
6:01
there's some extra lines that control
6:02
that I couldn't find any data on this uh
6:05
unit and the um their current product
6:07
range is quite significantly different
6:09
so we've got the uh blue these
6:13
might may well be saturating the camera
6:16
so they might look a bit weird so that's
6:17
the 430
6:18
blue the 575
6:24
yellow uh
6:26
475 light blue
6:29
the uh 520
6:31
green and the uh 630 red now one
6:36
interesting thing I noticed for the
6:39
575 it's actually it's actually using a
6:42
white lead and then filtering it rather
6:44
than using all the other ones are using
6:46
leads which are the fundamental colors
6:47
but uh this is actually doing white and
6:50
it's a combination of this filter and
6:52
the dichroic mirrors that are turning to
6:55
Yellow if we take the filter out and a
6:57
lot of the a lot of the um blue content
7:00
is going this way the red is going
7:02
straight through these two mirrors so
7:05
this is clearly not reflecting much of
7:08
that so we end up with the yellow coming
7:10
out of uh out of there which is a fairly
7:14
light yellow color which you don't
7:16
really see from high intensity leads so
7:19
that's clearly why they've used the
7:20
white to uh do this power consumption of
7:23
the white is pretty high so going up to
7:25
about 2 and 1 half amps on that color
7:27
whereas most of the other colors are
7:28
only drawing half an amp or so at 24
7:30
volts the uh the green is up to about
7:32
1.2 but say this thing is uh much
7:35
brighter and if you actually run all the
7:38
colors at the same time you get a fairly
7:41
reasonable um looking white coming out
7:43
of it and one thing you might just be
7:45
out to notice is there is some sort
7:46
color banding around here that's not
7:49
getting uh everything s completely
7:51
concentric and I think that's where this
7:53
fiber optic thing comes
7:58
in I'll
8:00
get a couple of Fairly accurately shaped
8:04
very sort of uniform color and looking
8:06
at What's um inside here we've basically
8:09
just got this Square Rod so this is
8:12
clearly yeah the lights just bouncing
8:13
off all the all the various sides to um
8:16
get a nice uniform illumination uh this
8:19
back bit looks like it's all potted so
8:21
nothing I really do to get in there I
8:24
think this is fiber so I have come
8:26
across um cables like this which are
8:27
liquid fill but just looking through the
8:30
end of this it's probably a bit hard to
8:31
see it does look like there fiber ends
8:34
going going on there and so there's this
8:36
feedback thing which is just obviously
8:39
compensating for the any light losses
8:41
through here to get an accurate
8:43
representation of uh the light that's
8:45
been launched out of these two
8:47
fibers and you see uh
8:49
these have got this sort of trapezium
8:54
shape light guides again it's like a
8:56
sort of acrylic or glass light guide
9:00
guess projected just to make the right
9:03
rectangular
9:04
shape and look at this Center assembly
9:07
um the light output doesn't uh change
9:10
whether you feed this in or not so it's
9:11
clear not doing any internal Clos Loop
9:14
control obviously there may well be some
9:16
facility for it to do that but it's not
9:17
being used in this
9:19
application and so this output just
9:21
produces a voltage on the uh outle
9:24
connector proportional to the amount of
9:26
light that's present so there's a little
9:28
diffuser in the back there
9:30
and then there's just some kind of uh
9:33
Optical sensor looks like a
9:35
chip looking at the lead it's a very
9:37
small package on the PCB with this lens
9:40
assembly over the top and these look
9:43
like they're actually on a copper
9:44
Metalized PCB for maximum thermal
9:47
performance and yeah it's a very small
9:49
package looks like it's a ceramic
9:51
package and there's a thermister there
9:53
for temperature monitoring this is the
9:56
475 blue one this is the 520 need to
9:59
Green which is uh rather different OB
10:02
it's a much bigger D with lots of bond
10:04
wise but also this looks like it's using
10:05
a phosphor if I shine a blue light at it
10:08
lights up green so this is actually a
10:10
phosphor conversion green lead which
10:12
I've I've come across before they want
10:15
that specific wavelength so they may be
10:17
easier to tune a phosphor than tune the
10:20
um semiconductor material to get the uh
10:23
right right wavelength from the lead
10:24
directly uh red 630 similar size to the
10:28
blue one or does seem to have a uh a
10:31
lens on top of it there is a sort of red
10:33
coloring to
10:35
the die but that doesn't appear to be
10:38
fluorescent as far as I can
10:39
tell and the white one again a little
10:41
bit different sort of much higher
10:43
current
10:46
connectors a makeer name on that
10:48
connector flot light not sure if that's
10:52
the connector or the lead
10:54
itself and obviously with the phosphor
10:56
and I'd imagine that phosphor may well
10:58
be tuned to get the maximum to the uh 5
11:01
cenm and actually this white one looks
11:04
like a St fairly standard product I just
11:06
found it in Mouse made by luminous
11:09
devices in fact actually I think all
11:11
these are based on various luminous
11:13
devices modules and they're you take
11:17
looks like they taking the nearest
11:18
wavelength and then just using these
11:19
filters to clean it up to get a precise
11:22
uh spectral line out of it so quite a
11:25
nice neat and um extreme
11:30
bright light source uh sure I've got any
11:33
particular use for it so I think this
11:35
might end up on
11:36
eBay but uh very pretty to look out and
11:40
without the uh risk of burning your eyes
11:43
out like you do with lasers so I thought
11:45
it would be interesting to try and
11:46
figure out the runtime of this things
11:48
like this we usually keep some sort
11:49
record of runtime cuz leads degrade over
11:51
time I couldn't get any software to work
11:52
through the USB face but then had a
11:54
thought probably going to be writing the
11:55
runtime periodically to the e s prom so
11:58
I just just scope up that and noticed it
12:00
was doing right every 5 minutes so I
12:02
just ran it for a while periodically
12:04
reading the E squ I just held the pick
12:05
in in reset and um put clip over to read
12:07
the square prom and found it was writing
12:10
one location per color every 5 minutes
12:12
so if one color was on it would write
12:14
that location every 5 minutes and just
12:16
increment it by one so after doing a few
12:18
tests with different colors of different
12:19
time periods it looked extremely
12:21
straightforward it's like a four bite
12:22
count for each color looking at the
12:24
original data that was in it all the
12:26
colors apart from Green were reading
12:28
zero and the green was reading four
12:30
indicating a total 20 minutes run time
12:32
ever if it was turned on run for a short
12:34
time then turned off that might not have
12:36
been counted but even so indicates this
12:37
thing wasn't used a great deal the whole
12:40
s process of doing a run can be several
12:42
hours but it'll only be doing probably
12:43
the Imaging at the end of that so you
12:46
wouldn't expect to be running for a long
12:47
time but say a single color for 20
12:50
minutes over its whole lifetime does
12:52
seem a little bit on the low side okay
12:55
let's look at the camera un fortunately
12:57
I managed to not record any sound when I
12:58
did this it's also a couple of months
13:00
ago so there's going to be a few details
13:02
that I've forgotten so I'm just going to
13:04
dub this over the original footage so um
13:07
take the lid off see this massive great
13:10
heat sink so this is a pel cool camera
13:12
we've got this blower fan producing a
13:14
fair amount of air flow through
13:16
it the connector here there's the ccds
13:19
mounted on the board on the
13:24
right this unplugs so we've got a bit of
13:27
power supply stuff on here
13:29
USB interface I think that's the Cyprus
13:32
microcontroller High speeded USB
13:34
interface there's a zyink spon fpga some
13:40
RAM and there's a couple of ATD
13:42
converters can't quite read what those
13:45
those are but anal
13:47
devices um little bit of bodgery around
13:51
here extra decoupling obviously they
13:53
have having some noise issues this is
13:55
around the ram chip quite a lot of extra
13:57
capacitors been added there
13:59
uh there's a couple of amplifiers prior
14:01
to the HD converter buffers or Andor
14:05
amplifiers taking the CCD
14:08
signal um bit more power spy stuff here
14:11
this is probably all to do with
14:12
generating the various CCD bias voltages
14:14
they uh need quite a lot of exotic
14:18
voltages next board down is just a
14:20
shield and an interconnect
14:24
boardly shielding the power supply stuff
14:26
from some the more sensitive an log
14:28
stuff
14:31
and this is the bottom board which is
14:32
just all power supply
14:34
stuff as you can see tons of capacitors
14:37
or Transformer in
14:42
there and this is the CCD which is a uh
14:47
very impressive thing this is a kf50 100
14:50
originally by true sense then codec
14:53
there ON
14:54
Semiconductor it's 50 megapixels uh the
14:58
only price I could find was this one
15:00
5,000 bucks and the architecture you can
15:03
see there actually two separate halves
15:04
which explains the Dual AZ converters
15:06
and two amplifiers it's literally split
15:08
down the middle and duplicated so it's
15:10
outputting two streams in parallel just
15:13
to keep the bandwidth sensible and it's
15:15
got this amazing um diffraction effects
15:18
it's got micro lenses over the pixel so
15:20
there's there's a bit more Optics going
15:22
on than on a normal
15:25
sensor few more bodges on the CCD board
15:28
including this wire which isn't really
15:29
tacked down very well which is a bit uh
15:32
bit of a mess quite a few bits around
15:34
this board where they've uh tacked
15:36
various bits on which is not super
15:38
impressive looks like CCD drivers on the
15:40
left with those 3 ohm um damping
15:43
resistors on the
15:47
output get a few more little bodges
15:50
around here some of
15:52
the and there's this separator the
15:54
silica gel to keep the moisture down but
15:56
there's this separator that actually
15:58
appears to be cut from piece of
15:59
antistatic
16:04
bag and this sort of thermal block on
16:06
top of this stack of three pel Cola
16:12
modules so as with any Stacks they get
16:16
um larger as they go back towards the
16:18
heat sink because each P's got to not
16:20
only take the heat from the previous but
16:21
also the waste heat which is quite
16:27
significant you see a little temperature
16:29
sensor here that copper block which
16:32
makes contact with the back of the
16:37
CCD and this's the back of the
16:40
pelas this then contacts the heat sink
16:44
on the uh rear there a few thermal pads
16:46
as well for some of the other power
16:47
components on this
16:51
PCB okay I've connected this uh camera
16:54
up I found some drivers on the disc that
16:56
seem to work under Windows 7 couldn't
16:58
get to install under Windows 11 though
17:01
um in the absence of any sort of lens or
17:03
being bothered to the proper amount I've
17:04
just put some f over it and put a little
17:06
pin in there to make a pinhole lens and
17:08
software gives a few options I'm not
17:11
entirely sure what all these are there's
17:12
obviously a clock frequency 22 MHz low
17:15
gain and with PFG no idea what that is
17:19
something something game programmable
17:20
Something game perhaps ver exposure
17:23
types I think focus is just like a
17:25
continuous grab until you tell it to
17:27
stop not entirely sure all these options
17:30
are obviously exposure time uh triggers
17:33
there ex external hardware trigger inut
17:35
you just trigger using a um thing on
17:37
screen so the resolution is 8176 by
17:40
6132 and you can actually bin those
17:42
where you combine multiple pixels to get
17:46
increased gain at the expense of lower
17:48
resolution down this is a 10sec exposure
17:51
obviously of the pin hole it's very uh
17:53
intensitive so we just stand still now
17:56
downloading it there's the uh exposure
17:59
so when it's
18:01
um there's a little status thing down
18:03
here so that tells you the um exposure
18:07
[Applause]
18:09
time it's this is just it
18:15
downloading um it is quite I'm seeing
18:18
quite a lot like smearing I think that I
18:20
don't know whether that's just due to
18:21
pixels overloading or something else I
18:24
mean yeah it's not it's not um out of
18:26
the question that there's something not
18:27
totally right about this camera
18:28
certainly was bodge wise on there um I
18:31
don't I'd imagine a camera like this
18:32
it's got a fairly narrow range of
18:34
intensities that it's happy with I'm not
18:36
going to spend a great deal of time on
18:38
this if you're interested in this camera
18:40
maybe for astronomy or something and
18:42
happy to sort of take the risk of it may
18:44
not be uh perfect I'll um I think I'll
18:47
stick this on eBay along with the
18:48
Illuminator I'll put a link down in the
18:50
description to the listing take your
18:52
chances to grab a bargain so for example
18:54
here we see this vertical streaking so
18:56
I'm not sure how normal that is this is
18:58
on fairly bright scene looking out the
19:02
window if I cut the exposure time down
19:04
on that it's now 1 second
19:07
exposure again most of the image
19:09
disappears again this is looks like it's
19:11
possibly over still overloading here go
19:14
that go down to say say quarter a
19:16
second so again I think there might be
19:19
some Auto gain control going on here um
19:21
this is with the PFG option let's try
19:23
turning that off and see what
19:25
happens so I'm not sure this is actually
19:27
more streaking or which just it's
19:29
cranked up the gain all the dis display
19:31
gray scale to show what um you know the
19:33
range of things that it's captured
19:36
there's one of one of 12 things in the
19:38
software there's um you can see of you
19:40
can't seem to read out the temperature
19:42
of the pelta cooler but you can set the
19:44
temperature and if you said it's a
19:46
different temperature you see the power
19:48
consumption jump up running the cooler
19:50
to get the temperature you requested but
19:52
I can't see anything anywhere that tells
19:54
you whether the cool is at the at the
19:56
temperature other than the power
19:57
consumption going down and there's no
19:59
temperature read out
20:03
here and just some yeah this is just
20:05
sort of very basic software I'm sure
20:07
there's like an API for more
20:09
sophisticated
20:10
applications but so if you know anything
20:12
more about these cameras please um stick
20:14
in the
20:15
comments um incidentally when I was
20:18
editing I didn't notice there was a bent
20:19
pin on the um CCD but I did fix that
20:22
before doing these tests and also
20:24
reactivated the um silica gel desicant
20:26
cuz I noticed it was uh I was getting
20:28
bit of condensation on the window but um
20:31
yeah so a couple of uh interesting but
20:34
maybe not particularly uh useful pieces
20:37
of Kit except for someone that's got a
20:38
very specific use so um I'll stick a
20:42
I'll stick these on eBay put a link in
20:44
the description and say hopefully
20:45
someone could actually make some uh good
20:47
use of these things
Example Output:
**Abstract:**
This video presents Part 2 of a teardown focusing on the optical components of a Fluidigm Polaris biotechnology instrument, specifically the multi-wavelength illuminator and the high-resolution CCD camera.
The Lumen Dynamics illuminator unit is examined in detail, revealing its construction using multiple high-power LEDs (430nm, 475nm, 520nm, 575nm, 630nm) combined via dichroic mirrors and filters. A square fiber optic rod is used to homogenize the light. A notable finding is the use of a phosphor-converted white LED filtered to achieve the 575nm output. The unit features simple TTL activation for each color, conduction cooling, and internal homogenization optics. Analysis of its EEPROM suggests extremely low operational runtime.
The camera module teardown showcases a 50 Megapixel ON Semiconductor KAF-50100 CCD sensor with micro-lenses, cooled by a multi-stage Peltier stack. The control electronics include an FPGA and a USB interface. Significant post-manufacturing modifications ("bodges") are observed on the camera's circuit boards. Basic functional testing using vendor software and a pinhole lens confirms image capture but reveals prominent vertical streaking artifacts, the cause of which remains uncertain (potential overload, readout artifact, or fault).
**Exploring the Fluidigm Polaris: A Detailed Look at its High-End Optics and Camera System**
* **0:00 High-End Optics:** The system utilizes heavy, high-quality lenses and mirrors for precise imaging, weighing around 4 kilos each.
* **0:49 Narrow Band Filters:** A filter wheel with five narrow band filters (488, 525, 570, 630, and 700 nm) ensures accurate fluorescence detection and rejection of excitation light.
* **2:01 Customizable Illumination:** The Lumen Dynamics light source offers five individually controllable LED wavelengths (430, 475, 520, 575, 630 nm) with varying power outputs. The 575nm yellow LED is uniquely achieved using a white LED with filtering.
* **3:45 TTL Control:** The light source is controlled via a simple TTL interface, enabling easy on/off switching for each LED color.
* **12:55 Sophisticated Camera:** The system includes a 50-megapixel Kodak KAI-50100 CCD camera with a Peltier cooling system for reduced noise.
* **14:54 High-Speed Data Transfer:** The camera features dual analog-to-digital converters to manage the high data throughput of the 50-megapixel sensor, which is effectively two 25-megapixel sensors operating in parallel.
* **18:11 Possible Issues:** The video creator noted some potential issues with the camera, including image smearing.
* **18:11 Limited Dynamic Range:** The camera's sensor has a limited dynamic range, making it potentially challenging to capture scenes with a wide range of brightness levels.
* **11:45 Low Runtime:** Internal data suggests the system has seen minimal usage, with only 20 minutes of recorded runtime for the green LED.
* **20:38 Availability on eBay:** Both the illuminator and camera are expected to be listed for sale on eBay.
Here is the real transcript. Please summarize it:
CH
20250509直播：习近平乱扯二战历史被台湾打脸；川普签第一份‘框架公告’，习普同日签下大单？习近平误判，用芬太尼摸到庙门；纽邮：川普最快下周把中国关税调至50%。（20250509第1839期）
大康有话说
241K subscribers
50,572 views Streamed live 7 hours ago
加入会员频道链接： / @dakang
PayPal账户：haikuotiankongnz@hotmail.com
联系信箱：dakangyouhuashuo@gmail.com
大康有话说频道网址： / 大康有话说
Transcript
Follow along using the transcript.
大康有话说
241K subscribers
Videos
About
Live chat replay
See what others said about this video while it was live.
60 Comments
wol pumba
Add a comment...
@alexchang5173
1 hour ago
對於台灣政治的評論，鏗鏘有力，完全認同。
7
@Roger007-n9n
1 hour ago
經過這次事件, 印度應該真正清醒過來, 它應該完全至少大面積站在美國一邊, 特別是現在全球正在選邊站, 應該成爲美國的盟友. 🤝
3
@wl9560
1 hour ago (edited)
台湾说二战后开罗宣言把台湾交给中华民国（当时就是全中国）：现在，中华民国=台湾，开罗宣言意味，如果中华人民共和国取代中华民国代表中国，开罗宣言就失取意义；或者，现在中华人民共和国和中华民国（即，台湾）是两个独立国家，所以开罗宣言=承认台湾独立，等等。没有意思：民进党要台湾独立，为嘛不敢堂堂正正说，开罗宣言胡说霸道，台湾原本就是独立国家。
@YRQ12345
1 hour ago
民國之前，中華大地5000年，只有夏、商、周，秦、漢、期間分裂成魏楚吳、南北朝、二晉和五代十國，及至陳橋兵變進入隋、唐、宋、元、明、清朝，從未有過一個”中國”；
@footprintonsnow
4 hours ago
中国这个名字，在法理上是不存在，要么PRC，要么ROC。被很巧妙地地混淆了。
6
@qwerty841120
1 hour ago
巴基斯坦引用的「擊墜」照片都被神通廣大的網友們查找出來是「墜機」的舊照片，而唯一一張近拍的照片也被YTR「美國叔叔」（他是原101空降師的上尉）打臉，理由有二：一是沒有任何的燒灼痕跡，二是上面那個非常明顯的標誌，他放大照片上的標誌後解說上面的編號，認為這個應該是幻象2000戰機拋棄的副油箱
@diqiuren76
1 hour ago
‘高度的尊重教宗’，就因为他反对川普支持开放边境吗？太牵强了吧。川普的很多政策不得人心，但严控边境。还是得到多数美国人支持的。
@gregoryyoung2347
3 hours ago
扭腰時報是民主黨吹鼓手，大康也是民主黨僱用的
6
@cchen185
2 hours ago
👍🏻🤗👏❤️💯
2
@jzchen6459
1 hour ago
欧洲应该对美国对等加税
@Cosmos-u3n
2 hours ago
賴清德隆重紀念歐戰勝利，推崇中華民國的貢獻。再一次確認台灣是中華民國的一省。這讓台獨前輩為之吐血。大康多做點功課吧！
@user-yp2ws8tc8v
1 hour ago
沒有綠營、兩岸緩和後、蔣經國也會解除戒嚴！選舉。 國民黨100多年來是民主政黨。只不過戰爭時期沒辦法。 總體來說是溫良恭儉讓、禮義廉恥的政黨！和綠營完全不一樣。
@yumlu2156
4 hours ago
中国有这么一个啥都不懂得当家人也是醉了。
3
@中森-j5g
4 hours ago
谢谢大康信息❤
4
@mokseuiyau
3 hours ago
👍👍👍大康
4
@cchen185
2 hours ago
大康老师，用Google查询什么牌子的电动车最不安全？竟然是tesla特斯拉！！！
@QuanwuLi
1 hour ago
你也別笑習近平被打臉，你也是在五十步笑百步。你不一直說中美貿易戰是習近平傲慢無視川普不斷遞來的橄榄枝，今天乍又說中國一直跟美國搭不上話？
@Poken.W
4 hours ago
CA$2.79
Thanks!
3
·
@rogerjsco
5 hours ago
马斯克捐款那么多，但是并没有得到什么照顾，川普的汽车关税对等关税对他来说就是伤害，他也和川普沟通过好几次也没有得到豁免或者其他好处，他捐款那么多价值在哪里？
第一，马斯克不是为了得好处而捐款，他为了和极左斗争不惜散财。第二，川普是为了美国，不为金主做事
13
@harryl8204
4 hours ago
大康以后像新闻联播一样播新闻就可以了，为了反川发表的左逼评论实在是反胃
12
Transcript
0:01
大家好 我是大康 今天是2025年5月9日星期五 开始我们今天的直播
0:10
我们今天节目的标题是 习近平乱扯二战历史最后被台湾打脸了
0:18
咱们看看这打脸是怎么打的 很有意思啊
0:23
另外呢 川普和英国最后宣布了 签了第一份的框架公告
0:30
这是从何说起啊 不是和英国签了第一份贸易协议吗
0:36
那还不能称为是贸易协议 后来彭博呢专门写了一篇东西 把这件事给解释了一下
0:44
我们看看他是怎么说的 再看看这所谓的框架公告里边
0:49
都有一些什么东西 那都是5月8号 川普在这搞了这么一个 也同样为了赶这个节日
0:59
他自己后来发帖也讲了 今天是欧洲的胜利日 也是美国的胜利日
1:05
你看我们又取得了这么一伟大胜利 这有点面子功成的意思 有点赶日子 那习近平和普京呢
1:14
正好在川普宣布第一份东西的时候 他们同日也签了十多份东西
1:22
那有俄罗斯媒体各方面报道 说他们也签下了大单
1:29
是吗 我呢查了一下双方的新闻稿 把这个大单
1:37
每次刚开始都有点别扭啊 把这个大单 我仔细的研究了一下
1:43
一会儿也跟大家说说 这大单都有什么
1:51
最后呢说这么一篇报道 说习近平误判 这是华尔街日报的报道 最后用芬太尼摸到了庙门
2:01
实际上习近平集团和川普团队 一直搭不上关系
2:07
马斯克那条线基本没用 最后怎么办呢 最后从头开始
2:13
中美两边从哪开始的 从芬太尼开始的 解铃还得系铃人
2:20
还得从头来 最后从芬太尼那儿 开始摸到庙门 开始进入状况
2:26
最后约了这个周末的瑞士会谈 那纽约邮报做了一个独家的报道
2:34
说川普最快下周 把中国的关税降到50%
2:41
从145降到50% 当然也有一说是降到50%到54%
2:50
54怎么说的呀 大板子上当时举的就是这个 34%加20是54
2:57
我更倾向于后边的一个数据了 咱们先听他这个题目
3:03
那好节目开始 还是先给大家介绍几个重要的国际新闻 我们还是先说台湾
3:10
先说赖清德 赖清德呢 昨天是5月8号
3:15
欧战的胜利日或者叫解放日 赖清德呢 他们也是第一次举办了纪念这个日子的一个活动
3:25
他邀请了欧盟以及17个国家的代表 大家都在总统府开了这么一个会 赖清德呢 做了这么一个讲话
3:36
后来他主要强调了三点 他说纪念这个日子有三个大的意义 第一个意义
3:44
二战胜利来自各国的团结抗敌 历史证明侵略注定失败
3:52
守护家园和自由的人 我们终将获得最后的胜利 这是第一个意义
3:58
第二 昔日的周星国就是当时我们联手抗纳粹抗敌的这些人
4:05
现在都变成了民主自由的繁荣之地 证明了只有自由民主才能带来发展和尊重
4:14
而威权独裁只能带来破坏与苦难 这是他第二个意义
4:20
第三 真正珍惜和平就不能对侵略者袖手旁观
4:27
说当前的我们和欧洲一样都面临着很多新型的威胁 你比如说切我们的海底电缆呢 等等 包括到选举的干预
4:37
经济的胁迫等等 说到自由阵营 我们必须要团结应对
4:45
当然这讲话要长得多了 这个呢 我只是把中间他说的这三个大的意义 三个要点给大家说说
4:54
我觉得讲挺好的啊 说来呢 这是台湾第一次在政府级别纪念二战的欧洲的胜利日
5:05
我认为这是一个非常好的开始 当然我今天已经听到有台湾的名嘴
5:11
啊 你看你赖清德啊 这个添美完了又添欧啊 或者怎么样
5:17
欧洲和我们有关系吗 离我们那么远和我们有什么直接关系啊 呃 不不不是这么一个情况啊 之前呢 我在节目里就呼吁过
5:28
说到台湾一直渴望得到国际承认是一个主权独立的国家
5:34
那你首先自己就要先做到 以一个独立国家的身份来参与国际事务
5:42
为国际的事务发声 参与这些国际大事 你就得行使一个国家的责任
5:49
你得做这些事 你呢 一直需要别人都来关心你来为你说话
5:55
但你有没有关心过别人为别人说话呢 为别人尽这些责任和义务的
6:03
说到这些关系都是互相的责任和义务也都是互相的
6:08
如果台湾自己以后在国际事务上 我就认为我自己是一个了不起的独立国家 我积极融入国际大家庭
6:20
我就照着这么做 我自己的分量 我做到那个份儿了
6:26
那别人想不承认你也难 那这一点我认为你自己首先要自要做到啊
6:34
如果像以前一样只是偏安一隅 然后等着别人来承认 那你就等着吧 那有日子吗
6:41
所以说原来在这一方面的胸怀和自信 我认为都稍差了一点 我认为现在到赖清德这算开始了
6:50
在这一点上为赖清德鼓掌 现在这个起点和胸怀和视野 我认为比以前又有所扩展
6:59
这个位置摆得更高了 我认为这一点做的是好的
7:04
那下面咱们就说说习近平 习近平又出什么事了 习近平到了俄罗斯之后 在那儿呢 在俄罗斯报上发了一篇文章
7:15
他去干嘛去了 纪念二战胜利日嘛 所以说他就说到了二战
7:20
他在里边当然就提到了台湾 他说台湾回归中国是二战胜利成果 和战后国际秩序的重要组成部分
7:34
开罗宣言 波斯坦公告都承认确认中国对台湾的主权
7:40
说到联大第2758号决议的权威性也不容挑战
7:45
不管台湾岛内形势如何变化 不管外部势力如何捣乱
7:51
中国是终将统一 统一的历史大势不可阻挡
7:57
他在里边提到了开罗宣言和波斯坦公告 都确认了中国对台湾的主权
8:06
他这个模糊着这么说 好像就确认了他对台湾的主权 是这么回事吗
8:12
结果台湾陆委会立刻就做出了回应 说1943年的开罗宣言和波斯坦公告 这些文件出现的时候
8:22
对不起 你这个中华人民共和国还不存在呢
8:29
你当时在哪呢 有你吗 台湾在二战之后确实被交给了中华民国
8:39
这也就是中华民国在台湾拥有合法地位的主要原因
8:44
台湾属于中华民国 你查那波斯坦公告开罗宣言 台湾都给谁了
8:52
给的是中华民国 那时候还没你呢 那时候你还是基因
8:59
就这么个意思 陆委会讲台湾从来就没有被交给过中华人民共和国
9:08
而且联大第二七五八号决议 美国这不是前两天给立了个法吗
9:13
也没有提到台湾 仅说那是驱逐蒋介石政权和代表中国的合法性
9:20
把蒋介石的代表给驱逐了 可没提到中华民国 当时是美国留了这么一个巧妙的后门
9:28
到今天你看人台湾陆委会这都知道啊 根本就没有讲到什么台湾的归属问题嘛
9:35
历史上从来就没有赋予中华人民共和国 管理代表台湾的权利
9:43
你在这乱扯什么波斯坦公告 开罗宣言 那时候你在哪呢 有你吗
9:50
你还在山沟里呢 所以说现在呢 人家讲从头到尾你也没管过我 你也管不着我
9:58
被台湾给打了这么个脸 所以说这点历史啊 这不能瞎说
10:04
一说啊 最后台湾 我估计大陆的很多朋友一听 你说当初我们习主席说的对啊
10:11
你说开罗宣言 波斯坦公告都是归我们吗 他就忘了一个事实 那时候有你吗
10:18
有这个中华人民共和国本朝吗
10:24
所以说说到呢 你原来中华民国对清朝是完整继承
10:31
房本过户是完整继承 连债务在条约在所有的东西全部继承
10:39
但是中华人民共和国对民国不是完整继承
10:44
说到第一次政协会上是有人提过啊 那不是我们也叫中华民国
10:49
那他那个就变成伪政府 我们因为我们占着北京占着中国呢 对吧 我们是正统啊
10:57
我们还尊着孙中山啊 当时就有人这么提过 当时毛泽东他们绝对得否
11:04
我怎么能供着孙中山呢 我这马恩列斯啊 对不对 那都是我的祖宗啊 那我不能供着老孙啊 那不可能的
11:13
所以说你是人家中华民国的国脉 人家的设计在台湾
11:22
人家是一个完整的继承过程 所以说到这儿 你说到台湾跟你有毛关系啊
11:28
你是个窃国者 你是个造反的 你是这么一个情况
11:33
为什么被联合国制裁了你那么多年 人家台湾一直在那坐着 只不过后来那美国总统一犯晕 那是另外的事了
11:43
来吧 这是给打了个脸 下边说说朱立伦 朱立伦呢 这几天疯狂的把赖清德比作希特勒啊
11:54
把这个说就说成是纳粹 说成是法西斯 然后一开始呢 昨天我们说到德国代表处人就不干了
12:02
你这不是瞎说吗 你这不是胡来吗 你动不动老拿我们这说什么呀
12:07
这是一回事吗 结果不光是德国说 后来比利时荷兰法国以色列
12:13
包括欧盟驻台代表处 多个国家和组织都转发
12:19
都指出了朱立伦的这个错误 最后这朱立伦呢 他不但不认错 要不然你就别说这事了
12:28
情绪还很激动 他专门呢 对这些西方国家针锋相对 他进行了反击
12:35
他讲外国政府不能成为迫害民主的帮凶
12:41
更不应该干涉他国内政 好家伙包括下边他有一些议员 包括立委啊 都认为这没毛病啊
12:51
你德国你给我闭嘴 你应该向骂我们党主席道歉 有人讲 甚至应该把德国代表处给他驱逐出去
12:58
好家伙就现在国民党疯了 就到这么一个地步啊 这事呢 你得仔细品
13:07
按说这么疯马牛不相及的硬操作 你这不是胡来吗
13:13
你为什么非要给赖清德 安上什么希特勒 法西斯纳粹这些称号呢
13:19
实际上你仔细品 这里边是有算计的 这里边是有深的算计的
13:27
说到一开始罢免国民党反罢免 他们提出的口号是什么呀
13:32
反绿共战独裁 四二六大游行 当时在凯达格兰大道举的都是这个啊
13:39
反绿共战独裁 结果这个绿共加上这独裁都触了中共的霉头
13:49
这么大一件事 中国国台办那边根本没报道 你为什么你这绿共你什么共 你骂我共
13:58
共因为绿共谁是主语啊 不共是主语吗 绿是形容词吗 你在骂共吗
14:05
你你独裁 你说谁独裁呢 你不都说我呢吗 所以说呢 当时共产党是很不高兴
14:12
那国民党那时候火速删改 反绿共战独裁就从那再也没有了
14:20
朱立伦就突然提出了所谓希特勒的这个系列的说法
14:26
到这我们可以想象 两边一定是沟通过的 那共产党肯定找国民党急了 你这什么意思啊 是你这胡说什么呢
14:35
你骂我 这两边一定是沟通过的 说到前面那什么 中共肯定很恼火了 不喜欢了 那下边说什么呀
14:45
我认为现在说的这套大概就是中共的意思了
14:52
朱立伦是美国的博士啊 是后来到了台湾的大学教授啊
14:58
你用这么生硬的不着边际的比喻 那实在是难以想象
15:03
太过牵强了 那你为什么要这么做呢
15:09
大家想普京打乌克兰的时候用的什么理由
15:15
我就是纳粹法西斯啊 我在这反纳粹化 打法西斯去了
15:21
这是普京师出有名的其中一个主要的名
15:27
那现在朱立伦在台湾内部 他就首先给提出来了 先给赖清德和民进党安上了什么 纳粹法西斯希特勒的帽子
15:38
那下边共军在进攻台湾的时候 这一条就跟俄罗斯一样
15:44
有那么三五条借口的话 这就是其中的一个主要的借口
15:49
你不但是台独 你在台湾岛内你还是希特勒 你还是纳粹还是法西斯
15:56
所以说到那我们就有合理性了 你这事你得这么想 你别看这么一个生硬的 不着边际的一个说法跑出来
16:06
这么多国抗议也死不改口 还跟人对呛 这可绝不是一个人的一时兴起 一时走嘴 这后边明显是有套路的
16:19
他是有一套说法的 所以说接下来我们来看 接着看说到这个说法
16:25
按说你已经引起了国际纠纷 台湾社会又比较在乎这个 而且你说的这个实在是不着边 你看他改吗 他不会改
16:37
因为我就认为 我这么体会 他就是领了对岸的圣旨 让他这么说的
16:44
这为下一步中共出兵 你看你们自己内部都说他是希特勒
16:49
你在那打压在野党 抄在野党的党部 他都来这个 所以说这是有阴险的算计的
16:58
这是我对这件事的体会 来下边说说韩国 那小金和老韩这俩事调整完了没有
17:07
昨天商量完还没调完 两边呢 我一看就后来有报道
17:13
我一看就谈不成 为什么呢 两边是公开谈
17:18
俩人弄一桌子 这边小金那边老韩 后边什么秘书啊什么乱七八糟站了一大堆人
17:25
坐在旁边围着 那怎么谈 那老韩就这句话 那咱俩合并那肯定是我呀 你让吧
17:38
那小金就不干了 说呢你老韩到今天 你也没有加入国民力量党
17:44
当然你是我们这头的 我们都知道 你一直刻意保持一个无党派的身份 所以说你不能像我们国民力量
17:53
正式通过初选的候选人来施压 你有什么权利跟我施压
17:58
我是正统你是野狐禅 两边呢就一直在那 咯咯唧唧在那说这个
18:06
两人谈后边一个大扇面 我看这有四五十人在旁边围着
18:11
这就没法谈了 这就没法谈了 最终呢 没有最终的结果
18:18
那小金呢就提议 金文珠就提议 那11号就是最后登记日了 再谈一次 再谈一次 能不能推迟到下周
18:30
他明确讲 如果这事处理不好 咱俩不能团结的话 我们俩都会被淘汰
18:36
所以说今天明天咱俩还得谈 所以说就这 你就没招 这边还没搞定呢
18:44
那边呢 李在明 人家那儿踏踏实实 来吧咱们说说印巴 先说最新的进展
18:51
又打了没有啊 又有一些最新的进展 印度政府说
18:58
说巴基斯坦的夜间 使用了无人机和导弹 发动了空袭
19:03
随后呢 印度方面采取了报复行动 他们摧毁了巴基斯坦那边 拉荷尔的防空系统
19:13
法新社也做了这个报道 说巴基斯坦的试图用无人机和导弹 攻击印度多个军事目标
19:21
最后印度国防部讲了 这些攻势都被我们的防空系统 所化解了
19:27
巴基斯坦那边倒没说什么 然后那边说 他们也打了我们的防空系统 就仅此而已了
19:34
还处在一种 有限范围的一个小规模的 这么一个状态
19:42
这是目前的一个最新的进展 那现在这罗生门 就是这战机 谁把谁打下来没有啊 这这这你得说这个呀
19:52
然后呢 印度官方做了一个报道 说在这个辛达行动中
19:57
印度，之前他可没说啊 这是最新的报道 我们击落了三架巴基斯坦战机
20:04
包括一架F16和两架骁龙17
20:09
我们打下你三个了 此外印度还在普尔瓦马 那儿有一个潘普尔地区
20:16
击落了一架试图 侵入印度领空的 骁龙17战斗机
20:27
那现在他们 人家倒没说 击落了什么歼10战斗机 这个他倒没说
20:33
他说呢我也打下你三架了 那要说到这个 确实也有凑数之嫌
20:42
印度呢也没有拿出任何的证据来 你看之前呢是那个是确凿的
20:48
巴基斯坦恐怖分子过去 跑那边杀了人家26个人 然后印度呢前两天打回来 用导弹攻击恐怖分子
20:57
最后巴基斯坦报了个数 一开始说是八个 还有俩孩子 最后呢给变成了26 明显有凑数之嫌
21:07
就是说这边杀了你26 然后你又打回来 我们也死了26
21:12
这就拉平了啊 这就差不多行了吧 你也出了气了 你也报了仇了 到此为止吧 他有这个意思
21:19
那现在印度说我也打了他三架战斗机 那什么意思呢
21:25
那说明前两天巴基斯坦报的 击落了三架印度战机 这事八成靠谱
21:34
你打了我三架 他心里自己知道 那我也说我也打了你三架 这样的话就拉平了嘛 咱俩不输不赢 是这么一个情况了
21:44
所以说我认为呢 这个也有明显的凑数之嫌
21:49
这里边大外宣环球时报 那这次的中国一片欢腾里边 他声挺大呀
21:56
他也报道了巴基斯坦击落了三架印度战机的消息 最后被印度驻华大使馆人家愤怒了 说你是散布虚假消息
22:06
说亲爱的环球时报 我们建议你在发布这样的虚假消息之前 能不能做一个事实核查和交叉比对
22:15
这是一个起码的做新闻媒体的一个责任 人家是挺不满的
22:23
那现在那就做做事实核查吧 昨天看到了画面上有那么多冒烟
22:28
印度战机在地下躺着呢 那不是证据吗 不是证据
22:35
那些都是老照片 最早的是2017年
22:41
印度军方确实发生了坠机的事故 最后就把那照片 中国这帮抖音包括网友们就给创造出来了 就给发出来了
22:52
然后最早的还有2005年的一些什么照片 那几个全是假的
22:58
因为你在那拍 巴基斯坦怎么会过来 因为他说都调到印度那边 你怎么得到现场的照片呢
23:06
印度那边因为这事不露脸 人家肯定严格控制各种消息来源
23:12
说印度找马斯克第一时间 你给我删8000多个号
23:17
都在推特上散布这些消息 印度直接告诉马斯克 你把这都给我删了
23:23
删了没有 马斯克绝对得听莫迪的 真删 所以说那些照片都是站不住脚的 中国这一片欢腾 那就是有病
23:36
那你说打下来没有啊 现在呢就出事了 这是路透社的独家报道
23:43
华盛顿这边给做了个确认 有几位美国官员表示
23:50
可以高度确认巴基斯坦使用中国制造的歼10战斗机
23:56
向印度战机发射了空对空导弹 击落了至少两架
24:02
然后被击落的这两架战斗机里 至少有一架是法国阵风
24:10
这是华盛顿证明了 然后法国人家也证明了 然后呢 那天是法国的一个官员向CNN也做了这样的透露
24:19
至少掉下一架阵风 这两位官员都讲巴基斯坦 这次并没有使用F16战机
24:28
所以说印度刚才说打下一架F16 那也是鬼扯了
24:33
人家根本没上 要上的话 美国是肯定知道了 大家注意啊 美国五角大楼的官员高度确认巴基斯坦使用歼10 打下你至少两架
24:47
那现在通过这种比对 我们大概能猜测 大概是打下三架来
24:55
然后印度这边说我也打下你三架了 他后来凑了这么个数 你从这个心态上 就跟两边在边界跟斗鸡似的在那踢腿
25:05
你踢一米五我也踢一米五 两边在那叫板 一样你说仨我也说仨 就这么个情况 这两个奇葩的国家
25:16
这是大概我目前的给我的这么一个印象了 那后边怎么办
25:23
巴基斯坦总理谢利夫强烈谴责印度 誓言报复每一滴血
25:30
印度防长辛格也说他的军队创造历史 行动尚未结束 嘿
25:37
看这两边都在这怼 但两边实际上在干嘛呢 都在找下台阶
25:44
这下台阶 一般来说呢 也就是俩国家应该是有这个下台阶的分量
25:53
其中一个当然就是美国 美国要是强力 你比如说国务卿现在就去了
25:59
先去巴基斯坦再去印度 两边这么一拍忽 这事基本就停了
26:05
你看那得了我们那美老大来了吗 那第二个能劝这件事的谁呢 不是中国
26:12
中国你可劝不了莫迪 莫迪可不吃你这一套 谁呢就是普京
26:18
普京在巴基斯坦和印度两边都有面子 那普京现在人家忙人家也没空 人家也不爱管你这事
26:26
5月9号你们都不去 那他才不管呢 那现在就是川普了
26:31
那川普呢 川普昨天的讲话里边呼吁和平 希望快速结束暴力 完了 就完了
26:41
他没有任何想要介入的意思 他不想管这件事
26:46
他只是表了个态 这美老大呢 他已经不像以前了啊 你要拜登在的时候 那这事肯定是严肃处理
26:55
川普政府是没这个爱好的 你们爱干嘛干嘛 这跟我我还费劲 还花飞机钱跑你那儿去 我没空 就这么个态度
27:06
那没有外力的强力组合 以这两个国家的智慧
27:11
那大概很难自己偃旗息鼓 会一直这么小打小闹
27:19
原本呢我认为甚至根本就打不起来 后来打起来呢 大家也都心知肚明 就是一场表演赛
27:28
没想到现在这是没有下台阶 现在就打成了一个小规模的消耗战
27:33
这个严格讲已经超出了双方一开始的预期了 尤其是印度现在有点骑虎难下
27:41
这时候如果就黑不提白不提了 这有点丢面子啊 和现在莫迪在世界上的这个排场 有点不符合
27:49
现在巴基斯坦人家无所谓 本来我们就灰头土脸的 印度现在莫迪现在面子大 那他丢不了这人
27:59
那会不会打大了呢 严格讲不太像了 我看是不太像了
28:05
咱们两边给比较一下 这双方已经不能严格讲算势均力敌了
28:13
印度现在是巴基斯坦人口的六倍 军费是巴基斯坦的八倍 GDP总值是巴基斯坦的十倍
28:23
外汇储备是巴基斯坦的48倍 特别是军力最近他因为有钱嘛 最近也添了不少的家当
28:32
两边的实力各方面完全不对等 巴基斯坦现在情况不好
28:38
高通胀 经济停滞 债台高筑 严重依赖中国掉入了中国的债务陷阱
28:45
是一个非常衰败的这么一个国家 而印度现在昂然向上
28:52
有点腾飞的意思 现在是繁荣发展 所以说这边比那边大十倍都多 严格讲你要说两边举国之力 要干一干
29:05
这个大家心里都有数了 来吧 下边说说乌克兰 乌克兰现在好不好
29:11
一会儿几个小时以后 那边就开始表演了 行不行啊 结果那边有报道
29:19
俄罗斯胜利日 肯定得有飞行表演队的表演嘛 结果这个训练基地 飞行表演队的基地 叫库宾卡空军基地
29:30
遭到了乌克兰无人机的袭击 这个基地就在莫斯科边上
29:36
是莫斯科勇士和雨燕 这两个飞行表演队的驻扎地
29:42
本来这就准备的 每天都挺忙的 在天上一块飞 结果袭击导致了基地 发生了猛烈的爆炸和火灾
29:53
那外界不知道 它具体的损失情况 因为它没有公布 谁也靠不了近 谁也不知道
30:00
这乌克兰真够硬的 他把这空中表演队 把他的老窝给掏了一下
30:06
当你是不是会影响 这次的这个阅兵的时候的飞行行动 这个现在不知道
30:13
这也算我们今天的一个看点吧 谈到阅兵的时候 咱们看啊 他那个勇士和宇艳飞起来没有 这也算是一个观察点了
30:23
说到呢 那俄罗斯联邦信息和通信技术监管局
30:31
他宣布了这么一个决定 断网
30:36
国家级断网 这已经不是莫斯科了 说这次的通信封锁覆盖整个俄罗斯60%的领土
30:46
俄罗斯可大呀 那老大的六成的国土 这次断网
30:53
俄罗斯的IT专家 就是他机构的专家证实 说这是俄罗斯历史上前所未有的一个案例 没发生过这样的事
31:04
俄罗斯联邦31个地区 已经事实上大规模断网
31:10
莫斯科和周边地区的达旦斯坦 巴什科尔托斯坦 什么萨玛拉
31:17
什么其他的这些 什么亚宾斯克 等等这些全部断网
31:23
你这一断网等于自动取款机了 出租车了 各方面谁也没法用
31:30
一下陷入到了混乱 就到这么一个程度 你说这办个阅兵就让人给吓成这样
31:38
说到今天 通知也发了 有人人家那能贴上来吗 莫斯科的市民都给贴上来了 说不能使用电梯
31:48
为什么呢 一旦被困 因为没有网 那信号也没法发出去 那操作员也不能知道
31:56
那你现在你困在电梯里 那你是有危险的 不能使用电梯
32:02
那这一断网 莫斯科机场还有数十架航班 航班要飞呢
32:08
莫斯科人都在那等着呢 等了多长时间 已经等了超过40个小时
32:14
说在叫福努科沃机场 8号到9号 大概有50多个出发和到达的航班都被取消了
32:23
然后在多莫捷多沃机场 大概有70多个航班被取消
32:28
然后还有在谢列梅捷沃机场 那也有70多个航班被取消了
32:36
人们在那骂 那你说怎么着 没网 那干脆谁也甭干了 就就就到了这么一个程度 我这是闻所未闻
32:46
那说到习近平去 中俄的这点子事 一会我们放在后边再说
32:52
那这场战争呢 川普发了个帖 说和俄罗斯乌克兰的谈判 实际上现在仍然在继续
33:01
他说作为美国 我们呼吁在理想情况下 双方应该达成一个30天的 无条件停火协议
33:08
希望双方能够达成可接受的 停火的条件
33:14
然后呢他讲我将追究 那些直接导致谈判破裂者的责任
33:21
如果谈判协议得不到遵守 美国和我们的伙伴将实施进一步的制裁
33:30
他和卢比奥两边的口吻不是太一样
33:36
卢比奥那边还是说 我们在一两周内判断 如果俄罗斯没这个意思 我们就要撤出
33:43
万斯呢也昨天在哪个场合 也流露了这个意思 说俄罗斯你就是故意不想谈 你那条件那么高怎么谈
33:51
说你要没诚意的话 我们就不管了 但是从川普的帖子里边 还没有这个意思
33:58
我希望你们达成 然后小泽他们现在没毛病 那就是你普京呗
34:03
不行的话 美国就要和合作伙伴 对你进行进一步的制裁 这个我们拭目以待
34:10
欧盟的外交部长卡拉斯 他讲呢 他欧盟各国的外交部长
34:16
一大帮外交部长 那你那边20多个国家领导人 普京的那帮小弟虾兵蟹将在那边去
34:26
那乌克兰显得挺冷清挺孤单 这卡拉斯带着各国的 欧盟的外交部长 5月9号我们一块去基辅
34:36
那去也不能空手去 宣布向乌克兰提供10亿欧元的援助 用于武器生产
34:44
然后呢 他们也会宣布 对俄罗斯的第17套制裁措施 所以说今天这边自由阵营 也呼啦啦一大波子人奔乌克兰
34:54
到那边来给小泽他们一块来助威了
35:01
这是乌克兰今天的一些最新进展 下边说说教宗
35:06
最新那烟通冒了白烟 经过那选举 从周三两天吧 这次算快的
35:16
133位枢机主教被封闭在 西斯廷礼拜教堂之内
35:22
最终以三分之二的多数 选出了教宗的继任者
35:27
这就是他的名字叫 罗伯特普雷沃斯特成为了新的教宗
35:34
这是来自美国芝加哥的 一个美国籍的教宗
35:40
那普雷沃斯特这位教宗 他是什么状态呢 现在一般看的他69岁啊
35:48
更多的细节我不给大家说了 他是教会里的一个改革派人物
35:55
那和原来的方济各呢 算是个继承关系 不是说职位继承了 是整个改革的继承
36:04
方济各呢 就是比较教会里的一个改革派 然后这位呢 也是啊
36:10
他跟方他原来就坚决支持方济各 所以说这是一个路线上的延续
36:17
最后出来啊 这个冒完白烟 大家欢呼 最后到了这个阳台上面对欢呼的人群 用意大利语表示祝大家平安
36:28
这位新教宗呢 会就可以非常娴熟的使用四门外语
36:37
这我我都忘了都什么语了啊 那意大利语是肯定是非常娴熟的了啊 他他语言是一个有有有语言天赋的 这么一个人了
36:47
那最后新教宗出来了 特别还是一个美国籍的教宗
36:52
一开始川普和万斯倒是念叨过 如果这次我们要出一个美国籍教宗多好呢
36:58
最后出来了以后啊 最后呢 这个万斯就祝贺 现在都有号了
37:03
叫列奥或者叫利奥十四世 他祝贺他当选为首位的美国教宗
37:11
当然现在有翻译还叫教皇 这个呢 教会已经很多年把这个皇字给去了啊
37:19
还是叫教宗 因为教皇 这还是有点过去中世纪啊 手握大权的那种 就是这等于是
37:29
教会和这个俗世的政权 好像有点分不清楚的那个
37:35
这这这还是有那个 现在人家早就改成叫教宗了啊
37:40
万斯讲的我相信数以百万计的 美国的天主教徒和其他基督徒都会为他祈祷
37:47
上帝保佑他 但是万斯呢 马上人家都在万斯的这句话下边都贴啊
37:54
贴一个 一条话 这什么呢 这位新任的教宗 列奥十四 利奥十四世
38:02
好像不怎么喜欢川普 也不喜欢万斯 之前他就有一个推特的公众号 他直接就批评万斯
38:12
直接就讲万斯你错了 说耶稣并没有要求我们对他人的爱来进行排序
38:21
这说什么呢 背后就对他们把他们分种族啊 等等这些东西啊
38:27
包括那什么性别了等等 所有人的爱没有先后啊
38:32
大家我们都一视同仁 他们是对他做了一个婉转的批评
38:38
那这么一个新的美国教宗出来 咱们就叫列奥十四世了
38:45
那他对目前川普的这个美国新政到底是一个什么态度呢
38:50
那毕竟他也是在美国出来的呀 那咱们就看看整个MAGA党的对这件事的反应
38:58
MAGA党两大悍将 叫男班农女就是那Lumer
39:06
那Lumer呢发了一个最新的推特 咱们看看这Lumer是怎么说的啊
39:12
说这就是新的教宗 他的名字叫罗伯特普雷沃斯特
39:18
他是第一位美国的教宗 他反对川普反对让美国再次伟大反对MAGA党运动
39:26
支持开放边境 而且和教宗方济各一样 是个彻头彻尾的马克思主义者
39:34
说天主教徒没什么好指望的 只不过是梵蒂冈又一个马克思主义傀儡罢了
39:43
男班农女Lumer 这Lumer这是MAGA党党的一个很厉害的人啊
39:48
他跟川普说他上话 川普经常没事还找他聊聊啊 一些重大活动他都去啊
39:56
他对这位教宗是这么看的 那这在整个就是MAGA党里边是有一定的代表性的啊
40:03
所以说呢 那上了一个美国教宗这一看都成他们家的了 那行不行啊
40:09
他不是一回事啊 这件事啊 对这位新任的教宗 我们要保持最高的尊重 这一点是毫无疑问的啊
40:18
来吧 那说说德国5月8号是欧洲的重大纪念日啊
40:24
那人们和以往一样重要的观察点观察谁呢 当然观察德国了
40:31
因为这5月8号是纳粹德国的战败日 是欧洲的解放日
40:37
那在这一天 德国人说什么呀 那别人哪都想说啥说啥 那把自己拔高都说
40:45
那当初的苦主德国就因为击败了你 我们就有了这一天
40:51
那他们怎么说的这些话呢 咱们看看啊 德国总统斯特伊麦尔在柏林议会发表了一个纪念讲话
41:00
看他怎么说的啊 对盟军士兵和抵抗运动表示感谢
41:07
称他们以全部力量付出了巨大牺牲 打败了纳粹德国
41:14
然后他也赞扬了苏联红军在击败纳粹中的作用
41:20
并且指出了啊 这个苏联红军从曾经解放了在纳粹占领下波兰的那个奥斯威辛集中营
41:28
说我们不会忘记这一点 历史你看德国的这个认知啊 还是一如既往的那么到位
41:39
你要指的日本的政客把话说到这个地步 我们感谢美国人来把当时的军国主义击败或者怎么样 啊
41:47
我们感谢当时中国的士兵啊 英勇奋战 哦 那那一你要日本人说这个呢 那他就没法混了啊
41:56
日本呢还没到这个程度 这是事实了 但是斯坦依迈尔在说完了历史之后就说到了现实
42:05
但是我们坚决反对克里姆林宫今天的谎言
42:11
说他们把三年对乌克兰的进攻描述为是对新纳粹的斗争
42:17
然后借用当时二战的遗产为今天他们的侵略行为来做辩护
42:24
他说对乌克兰的战争完全不是什么对抗法基斯主义斗争的延续
42:29
普京的侵略战争 你是对一个自由民主国家的攻击
42:34
和什么 当时反对纳粹暴政完全没有任何共同之处
42:40
他对这件事呢进行了猛烈的批判 所以说大家一定要看啊 朱立伦那边包藏祸心
42:48
为什么把赖清德和民进党 他们给安上什么纳粹希特勒 这完全就是给习近平手里递刀
42:55
让习近平采用普京对乌克兰同样的理由 同样的出发点
43:01
然后再篡改二战历史 然后就举着这个旗号去攻打台湾
43:07
可恶之极 就可以这么讲 非常的令人痛恨 可恶之极
43:14
我不知道台湾的那些名嘴们有没有认识到这一点 应该把他给戒破啊 这是非常令人愤恨
43:21
当然还是那句话 老有朋友留言吗 说大康你怎么现在一下就完全支持民进党 变成了给民进党摇旗呐喊的一个人
43:34
昨天我在节目里说了几句啊 为什么不呢 为什么不呢
43:41
台湾的自由民主有今天 我可以这么讲 就是民进党艰苦卓绝斗来的结果呀
43:49
当然这里边不完全是民进党人 这是以民进党为主的
43:55
包括还有一些外围的人士 大家前仆后继 不怕把牢底做穿 大家拼来的呀
44:02
当然是在斗争的过程中 小蒋公啊 包括后来的以李登辉为代表的国民党人士 做了一些妥协和让步
44:10
但是主要的推动力是谁啊 领着十万人一百万人上街的玩命的那帮人是谁啊
44:17
都是民进党人啊 为什么不能尊重他们啊 而反过来有一些啊 你看那绿的
44:24
你好好反省一下 你是不是中了共产党的认知作战了 在共产党的语境里边 这绿的就不是好人
44:32
民进党一听跟妖魔一般的存在 你好好读读台湾的近代史
44:38
当然我不能说民进党都对 他什么都好 但是在你现在看到的自由民主台湾
44:46
无论是言论自由和自由度在各方面都是亚洲第一 都是一个令我们所谓啊文化概念的华人 炎黄子孙为之骄傲的一个存在
44:58
主要就是这帮人拼命拼来的呀 为什么不能站在他们的立场上呢 这没毛病啊
45:05
这个今天咱们补充几句 这个可没毛病 来吧咱们说说以色列
45:12
耶路撒冷的报道 以色列国防军确认 一个由胡塞武装从也门发射的瞄准以色列的导弹又来了
45:21
结果在沙特阿拉伯的上空被击落 啊 他都坠毁了 没有 没到让以色列拦截的程度
45:31
胡塞呢再次公开表示 那你在阿曼和美国有没有一个协议啊 有啊
45:37
是有一个协议的 但是不包括对以色列的攻击
45:43
我跟美国呢 以后在红海是不是美国商团 我就算了
45:49
但是凡是和你以色列有关的 去以色列的或者从以色列出来的 我就见一个打一个
45:57
他还是那么一个情况啊 他认为对航运的袭击行动完全不会停止
46:03
所以说川普呢就抽身就走了 以色列日报的报道说 川普和内塔尼亚胡的关系已经跌入谷底
46:15
以色列总理对川普 包括川普对内塔尼亚胡
46:20
现在两边都用越来越失望来相互描述
46:27
从这边呢 川普突然和胡塞达成协议抽身而去
46:33
让整个以色列都大吃一惊 然后呢 那为什么呢 川普为什么突然这么做呢 他私下已经多次表示
46:43
对以色列呢在鼓动 鼓动谁 鼓动沃尔兹吗 鼓动小海 他们这帮人当时18个人 信号门通话
46:53
当时牵头的谁组的那个群呢 不就沃尔兹组的那个群吗 前一阵为什么把他给开了 他和内塔尼亚胡他们俩
47:02
算好了扣鼓动着美国去打胡赛 当时那个决议就是那么做的
47:07
然后美国捶胡赛 以色列去炸伊朗 当时他们的整个的战略安排是这么安排
47:16
那川普非常愤怒 对这个整个你们私下背着我 在这商量这个 他非常的不满
47:23
而且他也认为 以色列总理在推动加沙的计划上 我给了你那个未来的建设蓝图啊 你这个这么长时间都不行
47:31
他认为行动非常不利 那为什么最近行了 这不是川普拍桌子了吗 把沃尔兹也给开了
47:40
那现在内塔尼亚胡有决心了 那好我执行以前的那个任务
47:46
所以说呢 现在的两边关系 现在已经处得非常紧张了 那下边就看内塔尼亚胡在加沙这件事上 帮美国把这件事能办到什么程度
47:58
能不能啊 让老川乐一个 现在两边是非常不满了
48:04
来说说几条美国方面的新闻 第一条新闻的很有趣
48:10
我认为这是一叶而知秋 它是有代表性的
48:16
在新兴那提选市长
48:21
选市长呢蓝的啊 这里边咱们就别民主党这边的人 咱们就不说了
48:26
结果代表共和党出战的有一个人叫鲍曼
48:32
一开始的那边50多的民调 这边30多的民调 民主党领先
48:38
到选的后半期呢 万斯突然出来给这个共和党的鲍曼 站摇旗呐喊
48:46
然后他透露一信息 这是我弟 这是我弟弟
48:51
那你怎么叫万斯 他怎么叫鲍曼呢 我们是同母异父 我们一个妈
48:58
爹不一样 所以说姓的姓就不一样 那也是弟弟 一个娘胎里出来的 我们是亲弟弟
49:06
所以说呢他就站着 希望大家看在我的面子上去给大家投票 那好吧 最后投出来了
49:14
早早的人家民主党那边的宣读宣布胜选了 最后呢 民主党人一开始这鲍曼拿多少民调 30多的民调
49:24
最后他得了多少票呢 得了12.9% 他得了2800多票
49:31
而那边呢 那共民主党那位得了82.5%的选票 横扫
49:38
你不说还好点 原来你不说还有三十几的民调 你这一说 大伙一听 哦 原来那是你弟呀
49:46
还剩12% 大家就品这事啊 就说现在呢 他们在上面的100天也罢 或者是怎么样
49:54
现在在底下已经暗流涌动 发生了怎样的变化
49:59
这个呢 这是一个小的一个案例 一点一点的未来的各方面的离选举
50:06
你比如说到2026 这也不远了 到今年年底 明年的这个中期选举差不多就开始有样了
50:13
各种民调就开始出来了 大家呢 大概也就知道
50:18
现在真正的民意到底是什么了 那下边说是以美国人 那是令人肃然起敬了
50:27
老盖茨呢 是做了一个宣布 未来20年内 我将通过盖茨基金会 把所有的财富基本上都捐了
50:36
他说的是几乎所有的财富 那如果还活着的话 怎么也得给自己留点 他得生活嘛
50:44
用来拯救和改善全球各地人们的生活 也就是20年嘛 2025到2045
50:52
这个基金会就正式关闭 也就是在这20年内 我得把这2000多亿美元 我把它全花了
51:01
我把它全捐出去 他现在资助着什么癌症基金会啊 什么阿尔兹海
51:08
那个什么阿尔兹海默症啊 就老年痴呆什么的等等 大概八九个大的研究机构呢
51:16
世卫组织他都给很多钱啊 他都给很多钱 所以说 这很了不起
51:24
这个呢跟我们香港的发哥是有一拼了 有一拼啊
51:30
如果这个做到的话 那是现代历史上最大的一笔慈善捐款 他说了这么一番话
51:36
他用的是卡内基原来的一句话 说一个富裕离世的人将以耻辱收场
51:47
他说呢 人们在我去世的时候 关于我的评论会有很多的话了 但是我死的时候 我绝不会成为一个富裕的人
51:57
就刚才说 一个拥有巨大财富 这么死去的人 你是以耻辱在收场
52:04
他说呢 我要把这些财富都捐出去 最后死之前 我把这事都给办了
52:10
这个应该说 那这他有什么好表演的呀 他有什么要非要秀秀自己的情怀
52:20
他没必要啊 那以前人家那么多年的世界首富 走到世界各地
52:27
大家都吆三喝四啊 前后追捧他名利 人家都已经过了那个时候了
52:34
岁数也大了 前一阵因为这女色啊 跟太太离婚呢 有点小问题
52:39
那现在岁数大了 对这事也不是个事了 所以说到这时候 能给世界留下这么一个正面的形象 一个背影 很了不起
52:50
前首富这样的现首富呢 华盛顿邮报是做了一个独家的报道
52:55
他披露美国政府 现在正在向面临关税的国家来施压
53:01
要求他们都批准马斯克的星链的项目 原来啊 在和那帮谈判的时候 还夹带私货 还夹带着这星链项目呢
53:12
问题就你这星链 由美国政府来推 大家更不敢用 大家都知道乌克兰当时怎么折腾 泽连斯基他们
53:20
谁敢用你这玩意儿啊 东西是不错了 但问题是一旦拿你这个当战略投资 战略项目 政府要用
53:29
连意大利 梅洛尼是马斯克的朋友啊 他都不敢要 那别人谁敢要啊
53:35
所以说现首富还在干这种事儿 来吧 咱们说说贸易战吧
53:41
川普呢 签下了第一份框架公告
53:46
不是贸易协议 是框架公告 咱们先说新闻稿啊 说川普呢和英国首相斯塔莫 5月8号共同宣布
53:55
说美国和英国达成了贸易协议 还还叫贸易协议呢 主要的内容呢 是英国政府那边说 说美国对英国汽车的进口关税
54:05
从25%降到10% 钢铝关税降到零
54:11
当然是有一个数量 有一个配额的限制了 商务部长卢特尼克呢也讲 说英国将向美国放开多个市场准入
54:21
包括乙醇牛肉 机械 农产品 这些东西都敞开 同时呢 英国还准备买100亿美元的波音飞机
54:31
啊 这是整个这个协议大致的一些内容啊 根据呢 两边的协议 美国可以从英国出口到美国的商品中
54:41
因为有一个10%吗 那10%没免啊 要征收60亿美元的关税
54:48
同时呢 因为你给我敞开了那么多市场 我们对英国会多出口50亿
54:57
这是一个这是一个立竿见影的啊 直接的收入啊 大概就有这么多了
55:04
那两边这算是怎么一个回事呢 什么叫协议啊 互相呢 这是交换吧 交易吗
55:12
两边的交易等于是英国政府同意进口美国的食品和农产品方面 做出了大的让步
55:19
来换取美国降低对英国汽车出口的关税
55:24
原来25你给我降到10 我这儿呢 给你敞开农产品 所以说整个的这个呢
55:30
就为其他国家下面和美国的贸易协议打下了一个样板 留下了一个标杆啊
55:38
所以说这个意义呢 是啊 是这么一个情况 然后彭博社对这件事是做了一个很深刻的深入的一个报道
55:48
他们是把这件事完整的弄清楚了 一开始没弄清的时候 川普上过今天该买股票了
55:55
我们两边签第一个协议了 我取得了第一个伟大胜利 股市肯定涨了 一开始确实涨了
56:02
到最后呢 随着彭博路透他们的报道 人们弄清楚了这个协议的具体的内容
56:09
那个激动就退却了啊 最后呢 到晚上收盘的时候差不多把那条涨率都给吐回去了
56:17
英国那边基本没什么反应啊 英国伦敦市场也是世界三大之一啊
56:23
基本没有什么大的震动 所以说就说明整个的这个川普宣布的重大贸易协议 仔细看内容的话
56:33
实际上是什么 是一个面子工程
56:39
彭博首先讲 川普没有签署所谓 你比如说之前
56:44
这周早期 印度和英国不是签了一个自贸协定吗 川普是没有直接签这个的权利的
56:52
说这个权利是属于美国国会 美国国会必须批准任何的贸易协定
56:59
而这个过程肯定超过川普给的这90天的延缓期
57:04
所以说呢 川普是不能和别人来签这个的 签一些草案是可以了
57:10
这个呢 所以说啊 双方这个不是贸易协议 那说具体内容啊 这是彭博的评价啊
57:21
彭博首先认为这件事的象征意义 大于实际内容
57:26
说整个的框架呢 仅仅包括 他并不是一个全面的两国的新的贸易协议啊
57:34
框架只包括为美国出口提供更快的通关流程
57:40
限度内在限额内降低英国汽车钢铝产品的关税 在这个数之内啊 是刚才说的
57:48
一超过那个 你还要交25%加原来那2.527.5的汽车的关税
57:55
钢铝也一样 我给你一限额 这个里边是免税的 一超过的限额啊 其他那25还要生效
58:02
他说大量关键条款都等待后续来谈判
58:07
所以说专家把这个目前达成的这个框架称为是缺乏实质内容的框架条款
58:15
更多体现为政治上的仪式性的胜利 这是彭博的原话
58:22
那市场上的反应股市短线上涨 但投资者保持了谨慎
58:28
一开始一公告川普在那个他那个平平台上还进行了一些股东主要的板块普遍上涨
58:36
但后来弄清楚了这个以后人们发现呢啊 也就这样
58:42
所以说乐观情绪是有限的啊 市场也是这么认为的
58:47
那下边留下了很多大的争议的焦点 比如说数字税 药品的关税
58:53
食品的监管的标准 两边是不一样的
58:59
你美国的食品符合美国标准是去不了欧洲的 是去不了英国的
59:05
那那边是不是也应该跟美国那个统一 这次也没定啊 这都悬而未决
59:11
所以说英国呢继续对美国的科技公司收他的数字服务税啊 药品啊等等这个该干嘛干嘛啊
59:19
英国维持对美国农产品更严格的食品安全标准
59:25
所以说那些农产品去了能不能过这还不一定呢 这次没谈
59:31
然后他讲的信息也比较混乱 就也就是说具体内容很多呢都是比较粗糙的一些提纲
59:40
说双方的信息都是矛盾的 协议的可信度也遭到了质疑
59:46
你比如说英国说钢铝产品关税降为零 那边说呢只是一种限额
59:53
可是那个限额是多少呢 很多的细节还存在着分歧
59:58
彭博讲所谓的框架协议只是一个松散的政治姿态 是一个非具约束力的协议的文本
1:00:07
就是拉了一个大纲是这么一个东西 彭博接着写那他背后的动机是什么呢
1:00:14
川普非要赶到5月8号这一天宣布胜利
1:00:20
他主要是想缓解经济上的焦虑 正好他抢在这个欧战胜利80周年嘛
1:00:27
想借助这个重燃市场对他经济政策的信心
1:00:32
他最近民调支持率一度一直在下滑
1:00:38
就包括刚才说到万斯他弟弟选举啊 实际上他们作为当事人自己是很清楚的
1:00:44
川普是急需一项有卖点的外交成果 这是彭博的评论
1:00:52
那现在基本上可以看出呢 彭博下边这一条
1:00:58
10%没商量 连跟英国这样的铁哥们都没商量 那因为都知道世界整个的轴心英美
1:01:13
再往外亲戚往外扩展 那就是加拿大澳大利亚新西兰 我们都安格鲁萨克逊嘛
1:01:19
这五个是铁亲戚砸断了骨头都连着筋 这是核心
1:01:24
再往外才是日本才是欧盟才是这些国家 所以说跟最铁的里边的铁轴心都谈不下来 那别的就别商量了
1:01:36
这就死了心吧 这一镰刀肯定是没完 然后别的呢可谈
1:01:43
你看一个英国这不是也能谈下来吗 好多东西你不是 钢铝 这不给免了一个零吗
1:01:49
汽车一下给免了15% 你还怎么着啊 对吧 这就可以了
1:01:54
作为英国为什么要接 那不还有10%的吗 英国很踏实
1:01:59
因为你是什么 你是普涨 原来我的价格优势 因为大家一块涨 那我的价格优势还在
1:02:08
原来呢 你比如现在是这么一个情况 那大家都涨 那我原来那个市场份额 我的生态是没有受到任何打击的
1:02:17
那现在普涨的结果是什么 这10%是不动的了 川普想建这么一个水坝 干嘛呢 来保护自己国内的制造业
1:02:28
这样的话 你们的成本就比美国的制造业多了 至少10%
1:02:33
这还不到美国来投资 美国还制造业自己还起不来吗 我都给你把这个障碍壁垒都给你加上了 门槛都给你弄了 你还不行吗
1:02:44
那很多专家认为还是不行 10%还是练不过人家外边 打不过外边
1:02:51
还是这竞争力还是不够 那最终的结果 制造业也没回来 结果你这门槛全砸在美国人民的脑门上
1:03:00
最后美国人民所有的商品 至少10%的物价 这是大概要涨的了
1:03:05
这是目前的这么一个判断了 那现在看呢 不管怎么说 彭博讲英国是第一张多米诺骨牌
1:03:14
通过快速的有些粗糙的 有点抢时间的面子工程
1:03:20
他一定要把那标杆先立下 然后后边和印度日本韩国这些国家 我得铺路
1:03:27
当然最后和中国 这是整个这次的贸易战当中最艰难一战
1:03:34
这是彭博的报道 他最后的评价是这么说的 他说这是一个框架公告
1:03:41
不是一份协议 这是彭博结论性的这么一句话
1:03:50
那彭博专栏有作家也评论说 这是一个政治上的胜利 并不是经济上的突破
1:03:57
然后另外的一位专家讲 我们将看到更多这种框架 而不是协议的做法
1:04:05
也就是这种提纲 后边跟一国家一国家签的全是这玩意
1:04:10
剩下的具体细节谈判 怎么也得到今年年底了 都不一定谈完 所以说这是一个说法了
1:04:19
然后另外英国那边的大媒体讲 英国的金融市场反应冷淡
1:04:24
说明这个框架公告影响有限 我这是把整个彭博社 包括路透社
1:04:32
他们这些综合的报道 我给大家透露一下
1:04:37
大家也就知道了 这取得的第一个伟大胜利 它实际的分量到底是多大
1:04:45
它的成色到底是多大 这个我相信大家就清楚了
1:04:50
第一时间美国汽车制造商立刻是抨击
1:04:56
抨击谁 抨击川普和美和英国达成的协议 这里边有通用福特 这几家的大老板全急了
1:05:07
因为他们的重点在哪 在加拿大和墨西哥 他们的分厂基本在这两个国家 说你不先和这两个国家达成协议
1:05:17
先和英国达成协议 通用的老板就直接讲 根据这项协议
1:05:23
进口一辆几乎不含完全不含美国零件的英国汽车 加多少税才加10%
1:05:30
比从墨西哥或加拿大进口一辆符合美墨加协定 含有一半美国产零件的车都便宜
1:05:39
你这不是把我们给出卖了吗 你为什么对英国这样 这严重损害了美国汽车制造商 供应商和汽车工人的利益
1:05:50
美国汽车政策委员会的主席叫布伦特 也发出了这样的表述
1:05:56
这第一个反应就出来了 那再往后接着看 再往后接着看 这后边的事是有点意思
1:06:04
到这咱们评论几句 尽管刚才彭博包括几家媒体讲 大家后来整个的股票涨上去
1:06:11
后来市场看到了各家媒体的报道 一看就是个这 最后冷却 最后把那点涨幅又都吐回去了
1:06:21
这尽管这不是一个标准的贸易协议 算是一个献礼的工程
1:06:26
但不管怎么说 在整个这局贸易战里边 它有着强烈的象征意义
1:06:32
刚才讲了给世界定了个调 英国这么铁哥们 10%这别商量了
1:06:38
日本欧盟你们就不要再争争取了 欧盟这10%都不交
1:06:44
我多买你东西还不行吗 那美国说你多买我多少？ 我多买你500亿 第一轮就这么谈的吗
1:06:51
不成 美国人不答应 那我还要那10% 那欧盟今天把那单子都开出来了 什么塑料制品什么的等等的
1:07:00
1000多亿美元的东西 我给你也加上 会吗 我们都知道了 这是一种谈判 做张做势的一种姿态了
1:07:10
我个人的判断 这我不能笃定 因为毕竟乌克兰和俄罗斯那档子事没了
1:07:17
梅尔茨刚上来 又明确表示 他跟川普已经通完电话了 明确表示 美老大不能走 这一局里边必须还得有你
1:07:27
那未来欧洲还指着美国呢 你想在贸易谈判这 这个就谈不明白
1:07:33
所以我想英国都接了 那欧盟是不是考虑一下 我认为这也没啥选择了
1:07:40
目前看 在谈判端 总结这么几条 大概是这么一个结局
1:07:48
第一 10%的别商量了 这件事没商量
1:07:53
然后在接了这个的情况下 然后各国 他有一个呼应
1:07:58
也就是你打开多少市场 对我美国的食品 农产品 能源等等 你接我多少
1:08:07
我就会对你的钢铝汽车等等这些东西 我就会做出减少
1:08:14
取决于你对我开放多少 我就对你减少多少 所以说这是可谈的
1:08:21
那其他国家有很多国家 那你们都得全面削减 我美国出口的商品的关税
1:08:28
我到你那去 你现在先降低 你降低多少 我就跟你降低多少
1:08:33
这些东西都是可谈的 都是在10%的基础之上 然后加大的重点的力度 那就是川普的这俩大宝贝
1:08:45
一个是农产品 一个是农业能源产品 这两个是川普心心念念的
1:08:51
这两个你给川普面子 那在别的上面 我就会给你做更多的让步 这是一个这是什么了
1:08:59
另外还有一个重要的原则 就是你们得加大到美国来投资的力度
1:09:05
这也是川普盯着的事 这几个你们都满足 另外还有什么汇率控制 买我的国债 那都是往后排了
1:09:15
这几个你们只要都能满足我 那在10%之上 我给你加的那部分可以调整 可以谈
1:09:24
所以说下一步大家谈的基本就这么个情况 到现在可以宣布 英国的10%没拿下来
1:09:32
原来美国对全世界的关税是2.5 到现在已经变成12.5了 然后再加上对各国 那还得都搞点
1:09:40
那都不一定搞多少了 那要都这么搞下来的话 那美国的关税将是历史的 几乎是最高水平了
1:09:50
二战以后的最高水平 要维持相当长的时间
1:09:55
这韭菜割定了 所以说美国和世界之间的贸易的整个的角度
1:10:03
整个过去的秩序推倒重来 然后在完全基础要建立在新的门槛之上
1:10:10
这个新秩序就建立在这个之上 不讲理就不讲理 就这搁了
1:10:18
那还是那句话 你在高门槛内部 美国是自己是没有这10%的 那你这制造业能不能起来
1:10:27
这是川普的一个心愿 最后没起来 那最后美国老百姓白扛了几年的关税 那就是你的事
1:10:35
他能不能在2026中期选举中能说得过去 那就是那个事
1:10:40
所以说目前我们可以看得出来 欧盟大概没办法 真的是和美国开干吗
1:10:46
我到现在综合考量的话 我有点不太相信了 那日本韩国台湾什么的
1:10:53
欧盟都没戏 那你们更没戏了 所以说这个军事上都不自立
1:10:59
都指着美老大 这个没法谈 那下边还有谁
1:11:06
赌盘给开了这么一个 英国已经过了 然后下边是印度的赔率是85% 可能性是85% 就是印度了
1:11:17
日本是62% 越南是53% 怎么没韩国 韩国不着急 韩国还没选举呢
1:11:26
韩国大概得在6月中才能加入战团 这里边还没提台湾呢
1:11:31
因为印度那个是一个 因为他脚踩两只船 他又是金砖 他又是这边
1:11:37
所以说印度是一个单打的 一个另一个样板 英国这是铁哥们样板 这是给日本他们看的
1:11:44
印度是脚踩两只船的样板 和越南是打中国洗产地的样板 就看那条款怎么定
1:11:53
所以说他都选了各自的一个标杆 下边的这标杆一个一个的都立完了
1:11:59
你比如跟印度谈完了 后边还有别的什么脚踩两只船就好办了
1:12:04
因为越南洗产地的事一说 那后边什么马来西亚 菲律宾 印尼 这都顺理成章了
1:12:12
所以说这个呢 他是有各路的安排很有意思 目前看呢 因为中共所鼓动的大多数国家 我们团结起来围殴美国
1:12:23
这事呢不可能出现 结果现在都做不到联手 都是和美国单打
1:12:30
那谁也不是美国的对手 现在就上演了这个争先恐后和美国展开贸易谈判的景象
1:12:38
川普洋得意举着他那签字 他说这叫囚徒困境 是有这个了
1:12:45
抓起四个人来 你们一小团伙 我让你们单搁着 然后告诉你看那边都招了 你再不招你最后一个 我们可加重处罚
1:12:53
都吓唬一遍 结果这四个人争相 我说我说囚徒困境
1:12:59
这都到这么一个地步了 那现在落最后那谁 现在就是中国嘛
1:13:05
别人都跑了 而且都盼着我们态度好 他们想瓜分中国在美国的市场
1:13:13
瓜分在从中国跑出去的那些外资能上我们这来
1:13:18
都想跟美国谈一个好条件来吸引这些从中国出逃的外资
1:13:24
没说吗 这都不义气啊 都不仗义 就是这么个情况 那快速说说
1:13:30
那这边人家川普和英国这签了 那那边同样的时间都在五月八号 习近平和普京也签了一大堆东西
1:13:43
两边呢是签了一系列的会谈成果文件 两边呢也都讲了讲 也都讲了讲
1:13:51
那那边签大单 你跟普京也签大单 你这不是打对台打对台呢吗
1:13:57
你这不是较劲呢吗 嗯 我品了品两边的新闻稿 我也看了看双方的联合声明
1:14:05
咱们给大家分析一下 那普京和习近平他们俩搞成了些什么东西呢
1:14:11
咱们看看啊 这流氓之间能谈成些什么 人家那兄弟之间亲兄弟明算账搞成一这个
1:14:19
咱这看这俩黑老大之间弄成了一个什么 啊 这个联合声明老长了
1:14:25
我试着给大家念一下啊 中华人民共和国和俄罗斯联邦在纪念中国人民抗日战争 苏联伟大为国战争胜利和联合国成立八十年之际
1:14:36
关于进一步深化中俄新时代全面战略协作伙伴关系的联合声明
1:14:42
哈哈哈哈 这么老长的名字 我的天哪
1:14:47
那这联合声明里边有什么呢 主要达成了这么几个共识
1:14:52
政治和战略合作 这可以pass了 这这他们俩那点事啊 倒没说无上限了 但连我们俩结盟就完了嘛
1:15:01
关键是经济和贸易合作 双方同意扩大双边贸易规模 优化贸易结构
1:15:11
推动贸易和投资自由化便利化 扩大贸易规模 那在里边你们要干嘛呢
1:15:21
说到2030年大幅提升俄中双边贸易水平 就是在五年之内
1:15:27
目前双方外贸全都是用卢布和人民币来结算了 然后说我们这挺好
1:15:34
就说了个这 就说现在比如不到2000亿 我们慢慢把它给搞到2000亿 大概是这么一个情况
1:15:42
这是经贸 这是大家关注的 然后科技和创新 双方将加强在高新技术 人工智能数字经济等领域的合作
1:15:53
推动技术创新 都是干巴巴的 没有什么具体内容 然后说到人文交流 就是教育文化了
1:16:00
然后国际事务的协调 大概在联合国 他们叫联合投票等等
1:16:06
然后签署了20多个双边合作的文本
1:16:11
这个我们重点关注 20多个你们俩都签什么了 第一个全球战略稳定和国际法的权威
1:16:21
这个先放放 生物安全和检验
1:16:27
这也放放吧 投资保护和数字经济
1:16:32
文化和电影合作 20多个文本 20多个协议 就说了这四方面的东西
1:16:42
这叫什么呀 这让很多人应该说是感到有些意外的 有些意外
1:16:50
普京是说了一个 他提了一个 他要求中国把中国的生产设施 那些厂子 你能不能迁到俄罗斯了
1:17:01
那作为交换呢 我准备增加俄罗斯农产品对华出口
1:17:09
那这件事没有得到中国的支持 什么意思 等于你把你给我们你的工业基础 我给你们我们更多的粮食
1:17:22
那我怎么可能把我的工业基础给你呢 那你不是那过剩产能吗 你把你过剩的那些产能能不能给我
1:17:31
所以说你能看出这两个黑老大 我又想占你的便宜 我又不想呢 资敌心里互相还憋着劲呢
1:17:40
你比如说是习近平是普京的孝子了 那我把我的过剩产能给了你 你过几年起来了 我傻不傻 那不成
1:17:50
我还得压着你 所以说我拿着你那边的那好多高科技啊
1:17:55
发动机什么这些东西我也不能给你 两边的都在这斗着这个气呢
1:18:01
所以说呢到现在你刚才我给大家都念了 大家听出什么来了 这经贸合作各方面没有啊
1:18:09
另外所有人都关注的能源的 整个的联合声明和两边的谈话没提
1:18:17
那从外蒙古捅过来的那根大管子呢 没说呀
1:18:22
看不出这里边有什么具体的东西来 所以说那你说签什么大单了 刚才什么生物检验
1:18:30
什么国际法二十多项合作 所以说没看出有什么具体的一些实际的东西来啊
1:18:39
嗯 我品了一下 应该说习近平是留着余地的
1:18:45
最近和美国也接上头了 这谈判的陆陆续续就开始了
1:18:50
只要摸着庙门 只要摸到这个门把手的门已经开了一条缝 那后边就进去了嘛 后边就可以接着谈了
1:18:59
那习近平应该是做着准备 一旦和美国真的签约的话 那那第一阶段贸易协议 那两千亿美金多买两千亿美金
1:19:09
在2017年的基础上再加两千亿美金
1:19:14
你您想这事儿 那大规模的买买买大概是不可避免的了
1:19:19
为什么贝森特把那个拎出来 那是你签过字的
1:19:24
那是你以疫情为理由 你赖了账了 你可是咱们两边在团形办公室是签了字的
1:19:31
你不能说了不算啊 所以说那事儿人家还要捣后账 还要追你呢 而且你这么多年你都没买 你那个都不行 你还得赔我 你还得多买呢
1:19:41
所以说那你买美国什么呀 高科技又不卖你
1:19:47
你买那么多粮食 粮食单价又低又不值钱 那你说如果你的能源全买了俄罗斯的了
1:19:53
那美国的怎么办 你买美国什么呢 所以说现在呢 习近平应该说这次见到普京 是一个比较收敛的 模糊的这么一个谈法
1:20:05
大家呢查查新华社和俄罗斯那边的两边的报道 我相信呢 您大概和我会有同样的一个判断啊
1:20:13
习近平呢现在留着后手呢 那华尔街日报是登了这么一篇东西 因为习近平的误判导致局面升级
1:20:25
这和我们原来的说法是一致的 最后他用芬太尼终于摸到了庙门
1:20:31
把这个华尔街的这篇文章简单的给大家介绍一下
1:20:36
说现在呢 中国这就两边一直找不着途径啊
1:20:42
那怎么办呢 有知情人士说啊 跟华尔街日报说了 说中国向川普政府提出了一些问题
1:20:50
也就是试图弄清楚 你到底希望中国怎么来打击制造芬太尼的化学原料的贩运等等
1:20:59
你到底让我干些什么 他后来发现呢 因为所有的这些事儿 这两边这场架从哪打起来呢
1:21:07
从芬太尼打起来 加了两个百分之十 中国的态度当时相当的恶劣
1:21:14
因为确实那边芬太尼那商路没断 源源不断 那还有啊
1:21:20
川普认为这根都在你这儿 所以说加了一百分之十 中国回骂啊
1:21:25
你不能你 你咳嗽 让我吃药啊 最后你那你赖我 中国当时态度很强硬
1:21:32
又加了百分之十 当时又强硬 那后来解放日加百分之三十四啊 加起来就五十四啊
1:21:38
这边报复 然后两边都干起来了 是这么一个过程 那中国呢 现在就是解铃还得系铃人吗
1:21:46
他从头开始解 也就是咱们先从那这场架因芬太尼而起 咱们就从芬太尼开始
1:21:55
王小洪呢是问了美国 你们到底想让我们干嘛 白宫呢就向北京方面提供了一份建议的清单
1:22:04
我们要求你干嘛 然后两边的王小洪呢就立刻有来有回了
1:22:11
就立刻又和美国又进一步的接触 那我们想解决这问题 怎么解决
1:22:17
是我去美国还是去第三国啊 这两边呢正在接洽
1:22:23
那川普呢给他这个清单 就是你一定要向从事贩卖这些人员发出严厉的信息
1:22:31
这里边什么死刑啊 那些东西就肯定了啊 这这这都有了 人家提出了这个要求
1:22:37
然后 这个就进行了 就从这儿就两边就接上头了 就开始了
1:22:46
那别的呢 你芬太尼的事儿 这既然可以谈了 那贸易谈判的事儿呢 既然摸到了庙门 知情人士透露
1:22:54
然后说呢 贝森特和美国贸易代表格里尔和中国的对等官员和何立峰过去几天
1:23:03
同意将于本周晚些时候在瑞士会面 也就是从王小洪之后
1:23:09
何立峰和贝森特和格里尔一直在接触
1:23:15
一直在接触 这就是川普 为什么他有的时候 他可能不说实话
1:23:22
你能看出他来 但那时候如果就是张嘴就来 他完全不加思索 那应该是有诚实度的
1:23:29
他一直讲我们在谈着呢 然后中国说没有任何谈判
1:23:35
打川普脸 那川普就不高兴了 那你要是如何怎样 实际上过去几天 华尔街的报道一直在接触
1:23:44
最后才敲定这个周末 不像贝森特说的了 我们本来想去瑞士跟瑞士谈
1:23:51
碰巧发现中国人也在那里 顺便我们就谈谈 他就鬼扯了 这人家这把这事都给你整明白了 这事基本是这么开始的
1:24:02
和川普终于就说上话了 这件事是 川普是以芬太尼开始的
1:24:13
然后是中国官员当时认为 不会和因为芬太尼的问题 和美国进行接触
1:24:21
这是习近平的意思 中国官员认为 美国是以芬太尼问题为借口 对中国发起了贸易战
1:24:29
你是以这个为借口 到现在证明习近平领导层 这个态度
1:24:37
就当时人家芬太尼的时候 你又不去跟人配合 认为你是借题发挥
1:24:43
这是习近平的一个误判 这是华尔街日报的评论
1:24:48
说自那以后 川普就把对华关税涨到了145
1:24:55
就到了这么一个地步 所以说就因为这其实我们这个节目 一直也是呼吁习近平 他们还是去谈判
1:25:05
这种然后就上纲上线 鸡血上头什么下跪了 民族气节了
1:25:11
那扯什么呢 人家说芬太尼是不是一个事 你往人家美国投毒
1:25:18
你就跟当年鸦片战争一样 那不是一个性质吗 人家让你管哪儿不对了 人家没哪儿不对
1:25:26
然后你们就借题发挥 你那就是借口 然后你就不跟人家接触
1:25:31
最后怎么着 乖乖的按着人家一开始的路径 回到原点
1:25:38
从头跟人家美国人从头来 你当初何必 后边这一大套什么对等升级了 你扯这干嘛 打一开始
1:25:47
你为什么就不是这个态度呢？ 我们为这事可是在节目里 这个态度是一致的
1:25:54
华尔街日报他写了很多了 具体内容我就不给大家说 整个的梗概应该是这么一个状态了
1:26:03
那礼拜六要谈了 那怎么个谈法 下一步要怎么着
1:26:08
没想到纽约邮报 他给做了一个报道 最近纽约邮报可吃香
1:26:16
原来都是华盛顿邮报 什么纽约时报 什么华尔街日报什么的 原来都是在那儿 现在都让人给轰出去了
1:26:23
现在都是极右的一些媒体 在白宫 包括在五角大楼都吃香 都在前边
1:26:30
纽约邮报是现在跟川普内部的很多人 非常接近的一家媒体
1:26:37
他尽管是小开板的一个小报 但他的很多消息 现在看还是比较准的
1:26:43
他做了一个独家的报道 他说川普政府正在考虑一项计划
1:26:50
把对中国进口商品征收的145的关税 削减一半以上
1:26:57
最早将于下周生效 川普也讲了 如果和中国谈的好
1:27:03
也就是这个周六 两边的态度 就像习近平的话说 双向而行 是往一块谈
1:27:11
那好 他准备把这个调到一半 据接近谈判的消息人士透露 美国官员正在讨论这个提案
1:27:20
也就是把对中国商品的 收的惩罚性关税 降到50%到54%之间
1:27:29
然后在这个情况下 双方展开一场旷日持久的贸易谈判 来敲定一项贸易协议
1:27:36
也就是这个周末如果谈的好 先调到这儿 然后就带着这个关税运转
1:27:42
所谓的带病运转 在这个背着这个包袱的情况下 咱们继续谈
1:27:48
那就不知道谈多长时间了 也可能谈一年 也可能谈一年半 那对不起了 这50多的关税 你得一直背着
1:27:56
这像川普的第一场贸易战吗 非常像 川普那个也是 他后来签了贸易协议 我也不给你扯
1:28:04
那个关税也没拿下来 最后中国把它消化掉 通过转口把它给吸收掉 那是另外的事了
1:28:11
但这个54% 这也是不给了 不给了 然后对南亚邻国的贸易税 消息人士透露 可能会降到25%
1:28:23
这说谁呢 这说的就是越南 越南都40多 对这些国家可能要调到25%
1:28:32
然后他们讲 在整个的谈判期间 这关税大概就是最低50%了
1:28:40
那业内的人士讲 也是纽约邮报的报道
1:28:46
50%到54%的税率期间 是低于贝森特原来说的
1:28:52
不可持续的三位数的水平 然后和上个月川普和美国三大零售商老板会面 讨论的税率是一致的
1:29:02
现在沃尔玛什么的这些大家伙 甚至已经通知中国的出口商 你们可准备发货
1:29:11
现在都已经通知了 他这里边有沃尔玛 有Target 还有那家叫什么 叫加德宝
1:29:19
还有谁 好几家 这在白宫找川普去了 你这好家伙 这货轮都空着回来了 我们怎么办 这没法弄货架就空了
1:29:30
现在要把它降到这 川普应了他们了 我给你降到百分之五十几
1:29:35
这是我最大努力了 你们就承受吧 那哥几个一看 大概我们也行
1:29:41
现在就基本达成了这么一个 一个协议 等于川普是应了这零售三大巨头了
1:29:53
然后纽约邮报也讲 说我们收到的信号是 大坝将在本周末或者是下个周末决低
1:30:02
届时将会进行调整 也就是145 大坝要降低了
1:30:09
说很多的零售商已经开始要求供应商按照一系列的关税进行报价
1:30:16
也就是50%左右的来报价 这样他们就可以在商品抵达美国的时候来定价
1:30:23
就是先发货 发完货 咱们这东西在海上飘着 没到美国之前
1:30:29
我们这差不多就能搞定 到那时候咱们就能报出价来 就能进关了
1:30:34
现在就着急到这么一个份上 美国的市场就这么急 就到这么一个程度
1:30:40
今天西雅图的港口的老板 CEO接受福克斯专访 直接就讲我这没船了
1:30:49
我这看不见货船了 你说人家这慌不慌 我这好几千工人干嘛
1:30:55
我怎么办 谁能给我一个说法多长时间 现在纽约邮报爆这么一个大料应该就差不多了
1:31:05
当然他也白宫发言人也讲了 说关税决定将由总统亲自做出
1:31:10
其他的任何决定都是纯粹的猜测 他是这么一个说法了
1:31:16
然后说现在美国财政部的电话一直都处在被打爆的状态
1:31:22
白宫就威尔斯就办公室主任女士
1:31:27
接这帮什么零售商这帮的电话 也天天处在打爆的模式
1:31:34
上在那零售圈 现在的动作非常的迅速 非常的具体
1:31:41
这还没谈完呢 美国的零售商可能是川普透风给他们了 他们已经行动起来了
1:31:47
现在就已经到这么一个程度了 所以说大家我们心里也就有数
1:31:52
最后做一个评价 整个的这次周末的会面应该说不是无的放矢了
1:31:59
既然是约了这次的会面 在约之前应该说和立峰和贝森特他们是有所交流的
1:32:07
不可能是零基础 然后跑那空谈 那不成 应该是前面已经做了一些铺垫了
1:32:14
所以说他们心里是有数 那说到这个周末不可能有结果了
1:32:21
因为那和别人和英国和日本的情况不一样 中美之间是不可调和的 国运级别的矛盾
1:32:30
所以不可能高层见这么一次面 而且两边的以前都没见过面 就视频通过一次话
1:32:38
当时还非常紧张不欢而散 两个人没有任何的私人关系 更谈不上什么个人信任了
1:32:46
所以说这次就是接这么个头 大概就是三步了 第一步先破冰
1:32:53
然后把态度都表了原则定下来 我们还是要谈 我们两边还是要做 别脱钩
1:33:00
这基本就这点事儿 然后那好 我在那边关税给你降到50%多 你把我的报复关税你给我全撤了
1:33:09
我把那边报复关税也撤了 给你降到54% 这我们之前讲了很多次 他大板子就举着那个
1:33:17
就是54%了 然后背着咱们往下谈
1:33:22
建立一个对话机制 咱们往下谈 那第三步就到最后了 那就谈判具体的一些东西了
1:33:31
所以说这就是下边开始的一些内容
1:33:37
说到中国商品背着睡的已经抵达美国了 有一些商品已经在货架上有所展示了
1:33:46
有美国记者追查 这么看说 现在还没有对美国人的生活造成任何明显的影响 作为川普
1:33:55
那就是中国加税这件事 他已经开始他的软着陆了
1:34:00
因为美国市场的反应 老百姓的零售市场反应没那么强烈
1:34:06
如果要是未来要在两三周内和日本韩国这些国家都达成协议的话
1:34:12
那中国的滋味可就不好受了 所以说这个周末对中国来说四个字别无选择
1:34:22
川普讲 如果这个周末谈判进展顺利会降低关税
1:34:29
百分之五十几什么概念 这基本是川普在大选时候说的是多少 六十
1:34:36
他说了很多次了 这就是他们未来的一个标的 这基本上对中国
1:34:42
你看跟英国跟日本是百分之十 那跟中国大概五十以上 也就这样
1:34:50
这就是人家的底牌 这还没完 人家还得跟你谈汇率问题 跟你谈知识产权问题 跟你谈补贴问题
1:34:59
不行的话又上去了 所以说这个呢 只是给你个面子降到这个位置 然后跟你谈
1:35:06
一看谈不明白 那这又上去了 那就大脱钩了 所以说现在呢 这就这样了
1:35:14
这个就这样 那作为国内做生意的朋友们 你们就要体会了
1:35:21
大概是百分之五十到六十之间 就这个关税了 真够狠的
1:35:27
真够狠的 所以说你接还是不接 我认为在中国来说 四个字嘛
1:35:33
别无选择 你什么跪不跪 你扯那些都没用 上千万的工人没有工作
1:35:42
那么多的出口企业 也出现倒闭 你扯那些东西有什么用 对不对
1:35:49
现在呢跟你就是这么一个情况 你不自己做你的内需 不改善老百姓的生活
1:35:56
就全指的向全世界倾泻工业垃圾 最后收来外汇来发展自己的军备 发展自己的维稳力量
1:36:04
用那些钱准备打台湾造军舰 那人全世界谁也不答应啊 谁也不答应啊
1:36:10
所以说这次的贸易战呢 目前肉眼可见的 短期内是这么一个结局
1:36:19
那最终将来如何最终着陆 这个我们再拭目以待 现在还不好太说了
1:36:26
那好我们看看观众朋友们的观点 TTken
1:36:32
说台湾得重视这件事 给个期限 国民党 要么去大陆就别再回来了
1:36:38
要么就真的独裁一次 走新加坡的路线真给你看看 嗯
1:36:47
我分析的这个啊 我因为我前后的比照 从一开始啊反绿共战独裁
1:36:53
最后呢变成了什么希特勒 什么法西斯 明显是完全不合情理 完全是胡说八道 他为什么要这么说呀
1:37:02
哎呦 这可是往习近平手里给递刀 朱立伦要真干出这事来 而且我认为他就在干这件事
1:37:10
那么多国家人家都啊都都急了 他还在这顶 我认为他现在就是带着任务的啊
1:37:17
这帮玩意儿走到这么一个地步 那就是自绝于台湾社会 自绝于台湾人民
1:37:23
回去吧 赶紧滚蛋吧 小白公子说 印度的万国武器以后就不要得瑟了 各种系统是胡乱不堪
1:37:32
认准美国装备站边就别左右逢源了 可是问题是 以前他已经弄了那么多烂七八糟的东西
1:37:41
各个世代的 各个他跟他的卫星啊 跟他的雷达什么的都整合不到一起
1:37:49
要不说吗 枪有的时候是把好枪 你得分谁使
1:37:55
原来那些好的滑膛枪 当时给了中国 当时不是军阀混战吗
1:38:01
民国时期 最后就怎么用了那么多的子弹 最后也打不赢仗啊
1:38:06
后来有很多的西方军事专家实在不解 克鲁伯啊什么的那些大工厂就跑中国来了
1:38:13
结果呢 人家发现中国士兵怎么打 挖那沟一人多深 把这枪举到那沟上边
1:38:20
他人在沟里边 啪啪啪就这么打 把人鼻子都给气歪了
1:38:25
你这玩意能打着谁啊 你这不是吓唬人 你以为真拿这个放炮子呢 把这枪举过头顶 然后脑袋底下藏着这么打
1:38:34
是那枪是没毛病 你们怎么都这状态呢 那确实没办法啊
1:38:40
秉毅说中国要是没有美国市场 就等于40%的出口就没了
1:38:46
不光是这个和美国的直接贸易 关键是两边的对抗 就把所有的外资都吓跑了
1:38:54
因为现在跟这十几个大的贸易伙伴 那也是中国最大的贸易伙伴
1:38:59
欧盟这一把那就中国第二 那剩下的日本什么的 那都是跟中国有关的呀
1:39:06
那最后他来一个强迫站队 大伙一看你们俩这撕起来没完没了 最后在东海台海在擦枪走火 我们在中间夹着干嘛呢 我们撤吧
1:39:16
所以说中国的间接影响 实际上是没法想 细思极恐就到那么一个地步
1:39:25
吴家是闲谈梦落花 这有诗意呀
1:39:30
说别的地方哪有这么多韭菜呀 谈和最后协议是完和完全不同
1:39:36
这就是路透等权威大社和市场专业人士的专业判断
1:39:42
其实大家都在等着中国和美国之间的谈判结果 谈和拖是同时存在的事实
1:39:49
无论什么原因谈 双方如果都不想谈就不会谈 至于怎么开始的八卦 那个是无关紧要 只是个忽悠而已
1:39:58
那个呢因为中国是要这个脸面嘛 你看贝森特还讲鬼故事 给中国下台阶 那个是了
1:40:05
但是呢还是那句话啊 你得去谈去啊 你要斗争 你跟他当面斗争去 你不理人家
1:40:14
然后双方无限想象 都以极大的恶意来揣测对方 最后就无限升级
1:40:21
你不是吃饱撑的吗 所以说现在呢不管怎么说摸着门把手了
1:40:26
我认为后边呢 那也是真够丧权辱国的 背着50%以上的关税进行后边的谈判
1:40:37
情何以堪啊 怎么会到这个地步呢 那一开始人跟你说那俩10%芬太尼的时候
1:40:45
人家让你习近平跟他打电话的时候 那你为什么不跟人家打电话呢
1:40:50
你最后要背着54%的税再跟人家谈 这真是不是李鸿章嘛 这不是
1:40:58
詹姆斯如果川普如此摇摆 产业链是不会有丝毫改变 是如果你这个一天一变 谁敢制造业回到美国
1:41:07
因为投资那不是立即见效的 一两年以后可能才能看见模样
1:41:13
这是个长线 谁敢做长线 所以说这也是一个隐忧了
1:41:18
那好 那今天的节目就给大家做到这儿 谢谢这么多朋友的收看
1:41:24
希望大家都有一个美好的周末 我们在周日的老时间 再见
identifier: 2561
model: gemini-2.5-pro-exp-03-25| input-price: 1.25 output-price: 10.0 max-context-length: 128_000
host: 193.8.40.126
https://www.youtube.com/watch?v=yNaJkTnBojwinclude_comments: None
include_timestamps: 1
include_glossary: None
output_language: en
cost: 0.031415
*Abstract:*
This video provides a comprehensive overview of methods for evaluating Large Language Models (LLMs) on subjective language generation tasks, moving beyond simple benchmark scores. It categorizes evaluation techniques into three main types: reference-based evaluation (using metrics like BLEU, ROUGE, BERTScore, and COMET), human evaluation (including pairwise comparisons like LMSys Chatbot Arena and addressing biases with style control), and using LLMs themselves as judges (assessing outputs based on predefined criteria or pairwise preference, while also considering and mitigating potential biases). The video highlights the pros and cons of each approach, discussing aspects like cost, nuance, and the evolving role of AI in its own assessment, as exemplified by the MT-Bench findings where GPT-4 showed high agreement with human judgments.
*A Guide to Evaluating Large Language Models on Subjective Tasks*
* *00:00:02 Introduction to LLM Benchmarks & Evaluation Challenges:* The video begins by noting the increasing benchmark scores of new LLM releases (e.g., MMLU, GPQA for multiple choice; HumanEval for code). It highlights the difficulty in evaluating subjective tasks, such as explaining a concept clearly to a specific audience, and sets out to provide an overview of evaluation methods for such tasks.
* *00:00:52 Sponsor Segment: VoiceWriter-dot-io:* Introduces VoiceWriter, an AI-powered tool designed to help users write faster using voice. It features real-time grammar correction, punctuation addition, advanced speech recognition, and a Chrome extension. A free trial is offered.
* *00:01:22 Category 1: Reference-Based Evaluation:*
* This method is useful when a reference (ideal output) is available, common in tasks like summarization and machine translation. It's best for well-defined answers with limited variability.
* *00:01:52 BLEU and ROUGE:* These are traditional metrics that calculate the overlap of n-grams between the model's output and the reference. They perform surface-level matching and are not very nuanced, treating tokens as simply correct or incorrect without judging the degree of error.
* *00:02:39 Advanced Model-Based Reference Scoring:* More sophisticated approaches use language models.
* *00:02:45 BERTScore:* Employs an encoder LM to get embeddings for words in both candidate and reference sentences, then scores the similarity of these embedding sequences.
* *00:03:09 COMET and BLURT:* These metrics involve training a model to predict how a human would rate the similarity between two sentences. COMET, for example, encodes both sentences and uses a regression model for this prediction.
* *00:03:26 Category 2: Human Evaluation:*
* Involves humans directly assessing LLM outputs. Methods include rating output quality or, increasingly popular, pairwise comparison where users select the preferred of two model responses.
* *00:03:54 LMSys Chatbot Arena:* An open-source platform where users query two random models and pick the better response. Aggregated ratings generate an ELO score, creating a model leaderboard.
* *00:04:42 Pros and Cons:* While providing valuable insight, human evaluation is expensive and annotators aren't perfect, often influenced by superficial aspects like output length, formatting (e.g., use of bold text, emojis).
* *00:05:31 Solution - Style Control:* To address biases, a regression model can be fitted to account for style attributes (length, formatting) that might influence user preference, thereby disentangling style from true content quality. This showed some models (e.g., Claude 3.5) perform better after style control, while others (e.g., GPT-4o mini) perform worse, indicating their wins were sometimes due to formatting.
* *00:06:34 Category 3: LLM as a Judge:*
* This approach uses another LLM to evaluate the output(s) of one or more LLMs.
* *00:06:57 Setup:* The judge LLM can pick the better of two outputs or rate outputs on predefined criteria (e.g., helpfulness, relevance, accuracy, depth, creativity). Multi-factor rating provides deeper insights. (Example given: VoiceWriter uses LLMs to evaluate style adherence and content preservation for its grammar correction tool).
* *00:08:07 Validation:* The MT-Bench paper found that GPT-4's agreement with human judgments (85%) was even higher than human-human agreement (81%), suggesting LLMs can be reliable judges.
* *00:08:39 Potential Biases:* LLM judges can suffer from biases similar to humans, such as length bias (favoring longer responses), position bias (favoring the first response shown), or self-preference bias (a model rating its own outputs higher). These need to be controlled.
* *00:09:09 Cost and Utility:* While using LLMs for evaluation has a cost, it's typically cheaper than human evaluation, with a range of models offering different cost-quality trade-offs. LLMs are also useful for evaluating structured short answers where rule-based evaluation is hard (e.g., OmniMath benchmark for mathematical expressions).
* *00:09:53 Key Takeaway:* Evaluating LLMs on subjective tasks requires moving beyond simple metrics to a combination of reference-based, human, and LLM-as-judge evaluations. Each method has its strengths, weaknesses, and associated costs, and understanding biases (and how to control for them) is crucial for accurate assessment.
I used gemini-2.5-pro-exp-03-25| input-price: 1.25 output-price: 10.0 max-context-length: 128_000 on rocketrecap dot com to summarize the transcript.
Cost (if I didn't use the free tier): $0.03
Input tokens: 15148
Output tokens: 1248
Abstract:
This video provides a comprehensive overview of methods for evaluating Large Language Models (LLMs) on subjective language generation tasks, moving beyond simple benchmark scores. It categorizes evaluation techniques into three main types: reference-based evaluation (using metrics like BLEU, ROUGE, BERTScore, and COMET), human evaluation (including pairwise comparisons like LMSys Chatbot Arena and addressing biases with style control), and using LLMs themselves as judges (assessing outputs based on predefined criteria or pairwise preference, while also considering and mitigating potential biases). The video highlights the pros and cons of each approach, discussing aspects like cost, nuance, and the evolving role of AI in its own assessment, as exemplified by the MT-Bench findings where GPT-4 showed high agreement with human judgments.
A Guide to Evaluating Large Language Models on Subjective Tasks
Below, I will provide input for an example video (comprising of title, description, and transcript, in this order) and the corresponding abstract and summary I expect. Afterward, I will provide a new transcript that I want you to summarize in the same format.
**Please give an abstract of the transcript and then summarize the transcript in a self-contained bullet list format.** Include starting timestamps, important details and key takeaways.
Example Input:
Fluidigm Polaris Part 2- illuminator and camera
mikeselectricstuff
131K subscribers
Subscribed
369
Share
Download
Clip
Save
5,857 views Aug 26, 2024
Fluidigm Polaris part 1 : • Fluidigm Polaris (Part 1) - Biotech g...
Ebay listings: https://www.ebay.co.uk/usr/mikeselect...
Merch https://mikeselectricstuff.creator-sp...
Transcript
Follow along using the transcript.
Show transcript
mikeselectricstuff
131K subscribers
Videos
About
Support on Patreon
40 Comments
@robertwatsonbath
6 hours ago
Thanks Mike. Ooof! - with the level of bodgery going on around 15:48 I think shame would have made me do a board re spin, out of my own pocket if I had to.
1
Reply
@Muonium1
9 hours ago
The green LED looks different from the others and uses phosphor conversion because of the "green gap" problem where green InGaN emitters suffer efficiency droop at high currents. Phosphide based emitters don't start becoming efficient until around 600nm so also can't be used for high power green emitters. See the paper and plot by Matthias Auf der Maur in his 2015 paper on alloy fluctuations in InGaN as the cause of reduced external quantum efficiency at longer (green) wavelengths.
4
Reply
1 reply
@tafsirnahian669
10 hours ago (edited)
Can this be used as an astrophotography camera?
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
6 hours ago
Yes, but may need a shutter to avoid light during readout
Reply
@2010craggy
11 hours ago
Narrowband filters we use in Astronomy (Astrophotography) are sided- they work best passing light in one direction so I guess the arrows on the filter frames indicate which way round to install them in the filter wheel.
1
Reply
@vitukz
12 hours ago
A mate with Channel @extractions&ire could use it
2
Reply
@RobertGallop
19 hours ago
That LED module says it can go up to 28 amps!!! 21 amps for 100%. You should see what it does at 20 amps!
Reply
@Prophes0r
19 hours ago
I had an "Oh SHIT!" moment when I realized that the weird trapezoidal shape of that light guide was for keystone correction of the light source.
Very clever.
6
Reply
@OneBiOzZ
20 hours ago
given the cost of the CCD you think they could have run another PCB for it
9
Reply
@tekvax01
21 hours ago
$20 thousand dollars per minute of run time!
1
Reply
@tekvax01
22 hours ago
"We spared no expense!" John Hammond Jurassic Park.
*(that's why this thing costs the same as a 50-seat Greyhound Bus coach!)
Reply
@florianf4257
22 hours ago
The smearing on the image could be due to the fact that you don't use a shutter, so you see brighter stripes under bright areas of the image as you still iluminate these pixels while the sensor data ist shifted out towards the top. I experienced this effect back at university with a LN-Cooled CCD for Spectroscopy. The stripes disapeared as soon as you used the shutter instead of disabling it in the open position (but fokussing at 100ms integration time and continuous readout with a focal plane shutter isn't much fun).
12
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
12 hours ago
I didn't think of that, but makes sense
2
Reply
@douro20
22 hours ago (edited)
The red LED reminds me of one from Roithner Lasertechnik. I have a Symbol 2D scanner which uses two very bright LEDs from that company, one red and one red-orange. The red-orange is behind a lens which focuses it into an extremely narrow beam.
1
Reply
@RicoElectrico
23 hours ago
PFG is Pulse Flush Gate according to the datasheet.
Reply
@dcallan812
23 hours ago
Very interesting. 2x
Reply
@littleboot_
1 day ago
Cool interesting device
Reply
@dav1dbone
1 day ago
I've stripped large projectors, looks similar, wonder if some of those castings are a magnesium alloy?
Reply
@kevywevvy8833
1 day ago
ironic that some of those Phlatlight modules are used in some of the cheapest disco lights.
1
Reply
1 reply
@bill6255
1 day ago
Great vid - gets right into subject in title, its packed with information, wraps up quickly. Should get a YT award! imho
3
Reply
@JAKOB1977
1 day ago (edited)
The whole sensor module incl. a 5 grand 50mpix sensor for 49 £.. highest bid atm
Though also a limited CCD sensor, but for the right buyer its a steal at these relative low sums.
Architecture Full Frame CCD (Square Pixels)
Total Number of Pixels 8304 (H) × 6220 (V) = 51.6 Mp
Number of Effective Pixels 8208 (H) × 6164 (V) = 50.5 Mp
Number of Active Pixels 8176 (H) × 6132 (V) = 50.1 Mp
Pixel Size 6.0 m (H) × 6.0 m (V)
Active Image Size 49.1 mm (H) × 36.8 mm (V)
61.3 mm (Diagonal),
645 1.1x Optical Format
Aspect Ratio 4:3
Horizontal Outputs 4
Saturation Signal 40.3 ke−
Output Sensitivity 31 V/e−
Quantum Efficiency
KAF−50100−CAA
KAF−50100−AAA
KAF−50100−ABA (with Lens)
22%, 22%, 16% (Peak R, G, B)
25%
62%
Read Noise (f = 18 MHz) 12.5 e−
Dark Signal (T = 60°C) 42 pA/cm2
Dark Current Doubling Temperature 5.7°C
Dynamic Range (f = 18 MHz) 70.2 dB
Estimated Linear Dynamic Range
(f = 18 MHz)
69.3 dB
Charge Transfer Efficiency
Horizontal
Vertical
0.999995
0.999999
Blooming Protection
(4 ms Exposure Time)
800X Saturation Exposure
Maximum Date Rate 18 MHz
Package Ceramic PGA
Cover Glass MAR Coated, 2 Sides or
Clear Glass
Features
• TRUESENSE Transparent Gate Electrode
for High Sensitivity
• Ultra-High Resolution
• Board Dynamic Range
• Low Noise Architecture
• Large Active Imaging Area
Applications
• Digitization
• Mapping/Aerial
• Photography
• Scientific
Thx for the tear down Mike, always a joy
Reply
@martinalooksatthings
1 day ago
15:49 that is some great bodging on of caps, they really didn't want to respin that PCB huh
8
Reply
@RhythmGamer
1 day ago
Was depressed today and then a new mike video dropped and now I’m genuinely happy to get my tear down fix
1
Reply
@dine9093
1 day ago (edited)
Did you transfrom into Mr Blobby for a moment there?
2
Reply
@NickNorton
1 day ago
Thanks Mike. Your videos are always interesting.
5
Reply
@KeritechElectronics
1 day ago
Heavy optics indeed... Spare no expense, cost no object. Splendid build quality. The CCD is a thing of beauty!
1
Reply
@YSoreil
1 day ago
The pricing on that sensor is about right, I looked in to these many years ago when they were still in production since it's the only large sensor you could actually buy. Really cool to see one in the wild.
2
Reply
@snik2pl
1 day ago
That leds look like from led projector
Reply
@vincei4252
1 day ago
TDI = Time Domain Integration ?
1
Reply
@wolpumba4099
1 day ago (edited)
Maybe the camera should not be illuminated during readout.
From the datasheet of the sensor (Onsemi): saturation 40300 electrons, read noise 12.5 electrons per pixel @ 18MHz (quite bad). quantum efficiency 62% (if it has micro lenses), frame rate 1 Hz. lateral overflow drain to prevent blooming protects against 800x (factor increases linearly with exposure time) saturation exposure (32e6 electrons per pixel at 4ms exposure time), microlens has +/- 20 degree acceptance angle
i guess it would be good for astrophotography
4
Reply
@txm100
1 day ago (edited)
Babe wake up a new mikeselectricstuff has dropped!
9
Reply
@vincei4252
1 day ago
That looks like a finger-lakes filter wheel, however, for astronomy they'd never use such a large stepper.
1
Reply
@MRooodddvvv
1 day ago
yaaaaay ! more overcomplicated optical stuff !
4
Reply
1 reply
@NoPegs
1 day ago
He lives!
11
Reply
1 reply
Transcript
0:00
so I've stripped all the bits of the
0:01
optical system so basically we've got
0:03
the uh the camera
0:05
itself which is mounted on this uh very
0:09
complex
0:10
adjustment thing which obviously to set
0:13
you the various tilt and uh alignment
0:15
stuff then there's two of these massive
0:18
lenses I've taken one of these apart I
0:20
think there's something like about eight
0:22
or nine Optical elements in here these
0:25
don't seem to do a great deal in terms
0:26
of electr magnification they're obiously
0:28
just about getting the image to where it
0:29
uh where it needs to be just so that
0:33
goes like that then this Optical block I
0:36
originally thought this was made of some
0:37
s crazy heavy material but it's just
0:39
really the sum of all these Optical bits
0:41
are just ridiculously heavy those lenses
0:43
are about 4 kilos each and then there's
0:45
this very heavy very solid um piece that
0:47
goes in the middle and this is so this
0:49
is the filter wheel assembly with a
0:51
hilariously oversized steper
0:53
motor driving this wheel with these very
0:57
large narrow band filters so we've got
1:00
various different shades of uh
1:03
filters there five Al together that
1:06
one's actually just showing up a silver
1:07
that's actually a a red but fairly low
1:10
transmission orangey red blue green
1:15
there's an excess cover on this side so
1:16
the filters can be accessed and changed
1:19
without taking anything else apart even
1:21
this is like ridiculous it's like solid
1:23
aluminium this is just basically a cover
1:25
the actual wavelengths of these are um
1:27
488 525 570 630 and 700 NM not sure what
1:32
the suffix on that perhaps that's the uh
1:34
the width of the spectral line say these
1:37
are very narrow band filters most of
1:39
them are you very little light through
1:41
so it's still very tight narrow band to
1:43
match the um fluoresence of the dies
1:45
they're using in the biochemical process
1:48
and obviously to reject the light that's
1:49
being fired at it from that Illuminator
1:51
box and then there's a there's a second
1:53
one of these lenses then the actual sort
1:55
of samples below that so uh very serious
1:58
amount of very uh chunky heavy Optics
2:01
okay let's take a look at this light
2:02
source made by company Lumen Dynamics
2:04
who are now part of
2:06
excelitas self-contained unit power
2:08
connector USB and this which one of the
2:11
Cable Bundle said was a TTL interface
2:14
USB wasn't used in uh the fluid
2:17
application output here and I think this
2:19
is an input for um light feedback I
2:21
don't if it's regulated or just a measur
2:23
measurement facility and the uh fiber
2:27
assembly
2:29
Square Inlet there and then there's two
2:32
outputs which have uh lens assemblies
2:35
and this small one which goes back into
2:37
that small Port just Loops out of here
2:40
straight back in So on this side we've
2:42
got the electronics which look pretty
2:44
straightforward we've got a bit of power
2:45
supply stuff over here and we've got
2:48
separate drivers for each wavelength now
2:50
interesting this is clearly been very
2:52
specifically made for this application
2:54
you I was half expecting like say some
2:56
generic drivers that could be used for a
2:58
number of different things but actually
3:00
literally specified the exact wavelength
3:02
on the PCB there is provision here for
3:04
385 NM which isn't populated but this is
3:07
clearly been designed very specifically
3:09
so these four drivers look the same but
3:10
then there's two higher power ones for
3:12
575 and
3:14
520 a slightly bigger heat sink on this
3:16
575 section there a p 24 which is
3:20
providing USB interface USB isolator the
3:23
USB interface just presents as a comport
3:26
I did have a quick look but I didn't
3:27
actually get anything sensible um I did
3:29
dump the Pi code out and there's a few
3:31
you a few sort of commands that you
3:32
could see in text but I didn't actually
3:34
manage to get it working properly I
3:36
found some software for related version
3:38
but it didn't seem to want to talk to it
3:39
but um I say that wasn't used for the
3:41
original application it might be quite
3:42
interesting to get try and get the Run
3:44
hours count out of it and the TTL
3:46
interface looks fairly straightforward
3:48
we've got positions for six opto
3:50
isolators but only five five are
3:52
installed so that corresponds with the
3:54
unused thing so I think this hopefully
3:56
should be as simple as just providing a
3:57
ttrl signal for each color to uh enable
4:00
it a big heat sink here which is there I
4:03
think there's like a big S of metal
4:04
plate through the middle of this that
4:05
all the leads are mounted on the other
4:07
side so this is heat sinking it with a
4:09
air flow from a uh just a fan in here
4:13
obviously don't have the air flow
4:14
anywhere near the Optics so conduction
4:17
cool through to this plate that's then
4:18
uh air cooled got some pots which are
4:21
presumably power
4:22
adjustments okay let's take a look at
4:24
the other side which is uh much more
4:27
interesting see we've got some uh very
4:31
uh neatly Twisted cable assemblies there
4:35
a bunch of leads so we've got one here
4:37
475 up here 430 NM 630 575 and 520
4:44
filters and dcro mirrors a quick way to
4:48
see what's white is if we just shine
4:49
some white light through
4:51
here not sure how it is is to see on the
4:54
camera but shining white light we do
4:55
actually get a bit of red a bit of blue
4:57
some yellow here so the obstacle path
5:00
575 it goes sort of here bounces off
5:03
this mirror and goes out the 520 goes
5:07
sort of down here across here and up
5:09
there 630 goes basically straight
5:13
through
5:15
430 goes across there down there along
5:17
there and the 475 goes down here and
5:20
left this is the light sensing thing
5:22
think here there's just a um I think
5:24
there a photo diode or other sensor
5:26
haven't actually taken that off and
5:28
everything's fixed down to this chunk of
5:31
aluminium which acts as the heat
5:32
spreader that then conducts the heat to
5:33
the back side for the heat
5:35
sink and the actual lead packages all
5:38
look fairly similar except for this one
5:41
on the 575 which looks quite a bit more
5:44
substantial big spay
5:46
Terminals and the interface for this
5:48
turned out to be extremely simple it's
5:50
literally a 5V TTL level to enable each
5:54
color doesn't seem to be any tensity
5:56
control but there are some additional
5:58
pins on that connector that weren't used
5:59
in the through time thing so maybe
6:01
there's some extra lines that control
6:02
that I couldn't find any data on this uh
6:05
unit and the um their current product
6:07
range is quite significantly different
6:09
so we've got the uh blue these
6:13
might may well be saturating the camera
6:16
so they might look a bit weird so that's
6:17
the 430
6:18
blue the 575
6:24
yellow uh
6:26
475 light blue
6:29
the uh 520
6:31
green and the uh 630 red now one
6:36
interesting thing I noticed for the
6:39
575 it's actually it's actually using a
6:42
white lead and then filtering it rather
6:44
than using all the other ones are using
6:46
leads which are the fundamental colors
6:47
but uh this is actually doing white and
6:50
it's a combination of this filter and
6:52
the dichroic mirrors that are turning to
6:55
Yellow if we take the filter out and a
6:57
lot of the a lot of the um blue content
7:00
is going this way the red is going
7:02
straight through these two mirrors so
7:05
this is clearly not reflecting much of
7:08
that so we end up with the yellow coming
7:10
out of uh out of there which is a fairly
7:14
light yellow color which you don't
7:16
really see from high intensity leads so
7:19
that's clearly why they've used the
7:20
white to uh do this power consumption of
7:23
the white is pretty high so going up to
7:25
about 2 and 1 half amps on that color
7:27
whereas most of the other colors are
7:28
only drawing half an amp or so at 24
7:30
volts the uh the green is up to about
7:32
1.2 but say this thing is uh much
7:35
brighter and if you actually run all the
7:38
colors at the same time you get a fairly
7:41
reasonable um looking white coming out
7:43
of it and one thing you might just be
7:45
out to notice is there is some sort
7:46
color banding around here that's not
7:49
getting uh everything s completely
7:51
concentric and I think that's where this
7:53
fiber optic thing comes
7:58
in I'll
8:00
get a couple of Fairly accurately shaped
8:04
very sort of uniform color and looking
8:06
at What's um inside here we've basically
8:09
just got this Square Rod so this is
8:12
clearly yeah the lights just bouncing
8:13
off all the all the various sides to um
8:16
get a nice uniform illumination uh this
8:19
back bit looks like it's all potted so
8:21
nothing I really do to get in there I
8:24
think this is fiber so I have come
8:26
across um cables like this which are
8:27
liquid fill but just looking through the
8:30
end of this it's probably a bit hard to
8:31
see it does look like there fiber ends
8:34
going going on there and so there's this
8:36
feedback thing which is just obviously
8:39
compensating for the any light losses
8:41
through here to get an accurate
8:43
representation of uh the light that's
8:45
been launched out of these two
8:47
fibers and you see uh
8:49
these have got this sort of trapezium
8:54
shape light guides again it's like a
8:56
sort of acrylic or glass light guide
9:00
guess projected just to make the right
9:03
rectangular
9:04
shape and look at this Center assembly
9:07
um the light output doesn't uh change
9:10
whether you feed this in or not so it's
9:11
clear not doing any internal Clos Loop
9:14
control obviously there may well be some
9:16
facility for it to do that but it's not
9:17
being used in this
9:19
application and so this output just
9:21
produces a voltage on the uh outle
9:24
connector proportional to the amount of
9:26
light that's present so there's a little
9:28
diffuser in the back there
9:30
and then there's just some kind of uh
9:33
Optical sensor looks like a
9:35
chip looking at the lead it's a very
9:37
small package on the PCB with this lens
9:40
assembly over the top and these look
9:43
like they're actually on a copper
9:44
Metalized PCB for maximum thermal
9:47
performance and yeah it's a very small
9:49
package looks like it's a ceramic
9:51
package and there's a thermister there
9:53
for temperature monitoring this is the
9:56
475 blue one this is the 520 need to
9:59
Green which is uh rather different OB
10:02
it's a much bigger D with lots of bond
10:04
wise but also this looks like it's using
10:05
a phosphor if I shine a blue light at it
10:08
lights up green so this is actually a
10:10
phosphor conversion green lead which
10:12
I've I've come across before they want
10:15
that specific wavelength so they may be
10:17
easier to tune a phosphor than tune the
10:20
um semiconductor material to get the uh
10:23
right right wavelength from the lead
10:24
directly uh red 630 similar size to the
10:28
blue one or does seem to have a uh a
10:31
lens on top of it there is a sort of red
10:33
coloring to
10:35
the die but that doesn't appear to be
10:38
fluorescent as far as I can
10:39
tell and the white one again a little
10:41
bit different sort of much higher
10:43
current
10:46
connectors a makeer name on that
10:48
connector flot light not sure if that's
10:52
the connector or the lead
10:54
itself and obviously with the phosphor
10:56
and I'd imagine that phosphor may well
10:58
be tuned to get the maximum to the uh 5
11:01
cenm and actually this white one looks
11:04
like a St fairly standard product I just
11:06
found it in Mouse made by luminous
11:09
devices in fact actually I think all
11:11
these are based on various luminous
11:13
devices modules and they're you take
11:17
looks like they taking the nearest
11:18
wavelength and then just using these
11:19
filters to clean it up to get a precise
11:22
uh spectral line out of it so quite a
11:25
nice neat and um extreme
11:30
bright light source uh sure I've got any
11:33
particular use for it so I think this
11:35
might end up on
11:36
eBay but uh very pretty to look out and
11:40
without the uh risk of burning your eyes
11:43
out like you do with lasers so I thought
11:45
it would be interesting to try and
11:46
figure out the runtime of this things
11:48
like this we usually keep some sort
11:49
record of runtime cuz leads degrade over
11:51
time I couldn't get any software to work
11:52
through the USB face but then had a
11:54
thought probably going to be writing the
11:55
runtime periodically to the e s prom so
11:58
I just just scope up that and noticed it
12:00
was doing right every 5 minutes so I
12:02
just ran it for a while periodically
12:04
reading the E squ I just held the pick
12:05
in in reset and um put clip over to read
12:07
the square prom and found it was writing
12:10
one location per color every 5 minutes
12:12
so if one color was on it would write
12:14
that location every 5 minutes and just
12:16
increment it by one so after doing a few
12:18
tests with different colors of different
12:19
time periods it looked extremely
12:21
straightforward it's like a four bite
12:22
count for each color looking at the
12:24
original data that was in it all the
12:26
colors apart from Green were reading
12:28
zero and the green was reading four
12:30
indicating a total 20 minutes run time
12:32
ever if it was turned on run for a short
12:34
time then turned off that might not have
12:36
been counted but even so indicates this
12:37
thing wasn't used a great deal the whole
12:40
s process of doing a run can be several
12:42
hours but it'll only be doing probably
12:43
the Imaging at the end of that so you
12:46
wouldn't expect to be running for a long
12:47
time but say a single color for 20
12:50
minutes over its whole lifetime does
12:52
seem a little bit on the low side okay
12:55
let's look at the camera un fortunately
12:57
I managed to not record any sound when I
12:58
did this it's also a couple of months
13:00
ago so there's going to be a few details
13:02
that I've forgotten so I'm just going to
13:04
dub this over the original footage so um
13:07
take the lid off see this massive great
13:10
heat sink so this is a pel cool camera
13:12
we've got this blower fan producing a
13:14
fair amount of air flow through
13:16
it the connector here there's the ccds
13:19
mounted on the board on the
13:24
right this unplugs so we've got a bit of
13:27
power supply stuff on here
13:29
USB interface I think that's the Cyprus
13:32
microcontroller High speeded USB
13:34
interface there's a zyink spon fpga some
13:40
RAM and there's a couple of ATD
13:42
converters can't quite read what those
13:45
those are but anal
13:47
devices um little bit of bodgery around
13:51
here extra decoupling obviously they
13:53
have having some noise issues this is
13:55
around the ram chip quite a lot of extra
13:57
capacitors been added there
13:59
uh there's a couple of amplifiers prior
14:01
to the HD converter buffers or Andor
14:05
amplifiers taking the CCD
14:08
signal um bit more power spy stuff here
14:11
this is probably all to do with
14:12
generating the various CCD bias voltages
14:14
they uh need quite a lot of exotic
14:18
voltages next board down is just a
14:20
shield and an interconnect
14:24
boardly shielding the power supply stuff
14:26
from some the more sensitive an log
14:28
stuff
14:31
and this is the bottom board which is
14:32
just all power supply
14:34
stuff as you can see tons of capacitors
14:37
or Transformer in
14:42
there and this is the CCD which is a uh
14:47
very impressive thing this is a kf50 100
14:50
originally by true sense then codec
14:53
there ON
14:54
Semiconductor it's 50 megapixels uh the
14:58
only price I could find was this one
15:00
5,000 bucks and the architecture you can
15:03
see there actually two separate halves
15:04
which explains the Dual AZ converters
15:06
and two amplifiers it's literally split
15:08
down the middle and duplicated so it's
15:10
outputting two streams in parallel just
15:13
to keep the bandwidth sensible and it's
15:15
got this amazing um diffraction effects
15:18
it's got micro lenses over the pixel so
15:20
there's there's a bit more Optics going
15:22
on than on a normal
15:25
sensor few more bodges on the CCD board
15:28
including this wire which isn't really
15:29
tacked down very well which is a bit uh
15:32
bit of a mess quite a few bits around
15:34
this board where they've uh tacked
15:36
various bits on which is not super
15:38
impressive looks like CCD drivers on the
15:40
left with those 3 ohm um damping
15:43
resistors on the
15:47
output get a few more little bodges
15:50
around here some of
15:52
the and there's this separator the
15:54
silica gel to keep the moisture down but
15:56
there's this separator that actually
15:58
appears to be cut from piece of
15:59
antistatic
16:04
bag and this sort of thermal block on
16:06
top of this stack of three pel Cola
16:12
modules so as with any Stacks they get
16:16
um larger as they go back towards the
16:18
heat sink because each P's got to not
16:20
only take the heat from the previous but
16:21
also the waste heat which is quite
16:27
significant you see a little temperature
16:29
sensor here that copper block which
16:32
makes contact with the back of the
16:37
CCD and this's the back of the
16:40
pelas this then contacts the heat sink
16:44
on the uh rear there a few thermal pads
16:46
as well for some of the other power
16:47
components on this
16:51
PCB okay I've connected this uh camera
16:54
up I found some drivers on the disc that
16:56
seem to work under Windows 7 couldn't
16:58
get to install under Windows 11 though
17:01
um in the absence of any sort of lens or
17:03
being bothered to the proper amount I've
17:04
just put some f over it and put a little
17:06
pin in there to make a pinhole lens and
17:08
software gives a few options I'm not
17:11
entirely sure what all these are there's
17:12
obviously a clock frequency 22 MHz low
17:15
gain and with PFG no idea what that is
17:19
something something game programmable
17:20
Something game perhaps ver exposure
17:23
types I think focus is just like a
17:25
continuous grab until you tell it to
17:27
stop not entirely sure all these options
17:30
are obviously exposure time uh triggers
17:33
there ex external hardware trigger inut
17:35
you just trigger using a um thing on
17:37
screen so the resolution is 8176 by
17:40
6132 and you can actually bin those
17:42
where you combine multiple pixels to get
17:46
increased gain at the expense of lower
17:48
resolution down this is a 10sec exposure
17:51
obviously of the pin hole it's very uh
17:53
intensitive so we just stand still now
17:56
downloading it there's the uh exposure
17:59
so when it's
18:01
um there's a little status thing down
18:03
here so that tells you the um exposure
18:07
[Applause]
18:09
time it's this is just it
18:15
downloading um it is quite I'm seeing
18:18
quite a lot like smearing I think that I
18:20
don't know whether that's just due to
18:21
pixels overloading or something else I
18:24
mean yeah it's not it's not um out of
18:26
the question that there's something not
18:27
totally right about this camera
18:28
certainly was bodge wise on there um I
18:31
don't I'd imagine a camera like this
18:32
it's got a fairly narrow range of
18:34
intensities that it's happy with I'm not
18:36
going to spend a great deal of time on
18:38
this if you're interested in this camera
18:40
maybe for astronomy or something and
18:42
happy to sort of take the risk of it may
18:44
not be uh perfect I'll um I think I'll
18:47
stick this on eBay along with the
18:48
Illuminator I'll put a link down in the
18:50
description to the listing take your
18:52
chances to grab a bargain so for example
18:54
here we see this vertical streaking so
18:56
I'm not sure how normal that is this is
18:58
on fairly bright scene looking out the
19:02
window if I cut the exposure time down
19:04
on that it's now 1 second
19:07
exposure again most of the image
19:09
disappears again this is looks like it's
19:11
possibly over still overloading here go
19:14
that go down to say say quarter a
19:16
second so again I think there might be
19:19
some Auto gain control going on here um
19:21
this is with the PFG option let's try
19:23
turning that off and see what
19:25
happens so I'm not sure this is actually
19:27
more streaking or which just it's
19:29
cranked up the gain all the dis display
19:31
gray scale to show what um you know the
19:33
range of things that it's captured
19:36
there's one of one of 12 things in the
19:38
software there's um you can see of you
19:40
can't seem to read out the temperature
19:42
of the pelta cooler but you can set the
19:44
temperature and if you said it's a
19:46
different temperature you see the power
19:48
consumption jump up running the cooler
19:50
to get the temperature you requested but
19:52
I can't see anything anywhere that tells
19:54
you whether the cool is at the at the
19:56
temperature other than the power
19:57
consumption going down and there's no
19:59
temperature read out
20:03
here and just some yeah this is just
20:05
sort of very basic software I'm sure
20:07
there's like an API for more
20:09
sophisticated
20:10
applications but so if you know anything
20:12
more about these cameras please um stick
20:14
in the
20:15
comments um incidentally when I was
20:18
editing I didn't notice there was a bent
20:19
pin on the um CCD but I did fix that
20:22
before doing these tests and also
20:24
reactivated the um silica gel desicant
20:26
cuz I noticed it was uh I was getting
20:28
bit of condensation on the window but um
20:31
yeah so a couple of uh interesting but
20:34
maybe not particularly uh useful pieces
20:37
of Kit except for someone that's got a
20:38
very specific use so um I'll stick a
20:42
I'll stick these on eBay put a link in
20:44
the description and say hopefully
20:45
someone could actually make some uh good
20:47
use of these things
Example Output:
**Abstract:**
This video presents Part 2 of a teardown focusing on the optical components of a Fluidigm Polaris biotechnology instrument, specifically the multi-wavelength illuminator and the high-resolution CCD camera.
The Lumen Dynamics illuminator unit is examined in detail, revealing its construction using multiple high-power LEDs (430nm, 475nm, 520nm, 575nm, 630nm) combined via dichroic mirrors and filters. A square fiber optic rod is used to homogenize the light. A notable finding is the use of a phosphor-converted white LED filtered to achieve the 575nm output. The unit features simple TTL activation for each color, conduction cooling, and internal homogenization optics. Analysis of its EEPROM suggests extremely low operational runtime.
The camera module teardown showcases a 50 Megapixel ON Semiconductor KAF-50100 CCD sensor with micro-lenses, cooled by a multi-stage Peltier stack. The control electronics include an FPGA and a USB interface. Significant post-manufacturing modifications ("bodges") are observed on the camera's circuit boards. Basic functional testing using vendor software and a pinhole lens confirms image capture but reveals prominent vertical streaking artifacts, the cause of which remains uncertain (potential overload, readout artifact, or fault).
**Exploring the Fluidigm Polaris: A Detailed Look at its High-End Optics and Camera System**
* **0:00 High-End Optics:** The system utilizes heavy, high-quality lenses and mirrors for precise imaging, weighing around 4 kilos each.
* **0:49 Narrow Band Filters:** A filter wheel with five narrow band filters (488, 525, 570, 630, and 700 nm) ensures accurate fluorescence detection and rejection of excitation light.
* **2:01 Customizable Illumination:** The Lumen Dynamics light source offers five individually controllable LED wavelengths (430, 475, 520, 575, 630 nm) with varying power outputs. The 575nm yellow LED is uniquely achieved using a white LED with filtering.
* **3:45 TTL Control:** The light source is controlled via a simple TTL interface, enabling easy on/off switching for each LED color.
* **12:55 Sophisticated Camera:** The system includes a 50-megapixel Kodak KAI-50100 CCD camera with a Peltier cooling system for reduced noise.
* **14:54 High-Speed Data Transfer:** The camera features dual analog-to-digital converters to manage the high data throughput of the 50-megapixel sensor, which is effectively two 25-megapixel sensors operating in parallel.
* **18:11 Possible Issues:** The video creator noted some potential issues with the camera, including image smearing.
* **18:11 Limited Dynamic Range:** The camera's sensor has a limited dynamic range, making it potentially challenging to capture scenes with a wide range of brightness levels.
* **11:45 Low Runtime:** Internal data suggests the system has seen minimal usage, with only 20 minutes of recorded runtime for the green LED.
* **20:38 Availability on eBay:** Both the illuminator and camera are expected to be listed for sale on eBay.
Here is the real transcript. Please summarize it:
00:00:02 Every new model release comes with
00:00:02 higher and higher scores on various
00:00:04 benchmarks. How do these benchmarks
00:00:06 work? Many of them like MMLU and GPQA
00:00:10 are multiple choice questions where the
00:00:12 model is presented with four choices and
00:00:14 has to pick the right one. Some
00:00:16 benchmarks involve writing code like
00:00:18 human eval and live codebench. And to
00:00:21 evaluate these, you simply have to run
00:00:23 the code and see if the model passes all
00:00:25 the tests. Where it gets more
00:00:27 interesting is when the result is more
00:00:29 subjective like explain self attention
00:00:31 to a fifth grader. How do you even
00:00:34 measure the quality of this explanation
00:00:36 and how do you know if your model is
00:00:37 improving across time? My name is B. I'm
00:00:40 a machine learning engineer and a PhD in
00:00:42 natural language processing. In this
00:00:44 video, I will give an overview of all
00:00:45 the ways you can evaluate your large
00:00:47 language model on subjective language
00:00:50 generation tasks. Before we continue,
00:00:52 let me share something I've been working
00:00:54 on. Voice Writer is a tool to help you
00:00:56 write faster using voice. Simply speak
00:00:59 your thoughts and the AI will fix your
00:01:01 grammar and add punctuation in real
00:01:03 time. It uses advanced speech
00:01:06 recognition models and I use it every
00:01:08 day to write things much faster,
00:01:10 including my emails and my daily standup
00:01:12 updates. You can also use our Chrome
00:01:14 extension to use VoiceWriter on any
00:01:17 website. Try it for free at
00:01:20 voicewriter.io. Now, back to the video.
00:01:22 The first category of evaluation metrics
00:01:24 I'm going to go into is reference based
00:01:27 evaluation. This is useful when you have
00:01:29 a reference available or what the model
00:01:31 is supposed to output. Examples where
00:01:33 this is popular are summarization and
00:01:36 machine translation. But reference based
00:01:38 evaluation can be used for any task. It
00:01:40 is most suitable when the answer is
00:01:42 relatively well defined and there's not
00:01:45 too much variation in other possible
00:01:47 correct answers. and is less useful when
00:01:50 the answer is something creative. Two of
00:01:52 the oldest and most classic reference
00:01:55 based evaluation metrics are blue and
00:01:57 root. Given a reference and a candidate
00:02:00 sentence, the blue score basically
00:02:01 computes how much does the engrams in
00:02:04 these sentences overlap with each other.
00:02:06 The root score works in a similar way
00:02:09 and there are a lot of variations in
00:02:11 both of these metrics that I'm not going
00:02:12 to go into today. But in both of them,
00:02:15 the algorithm only looks at the surface
00:02:17 level a matching and is not very
00:02:19 nuanced, like every token is correct or
00:02:21 incorrect. So anytime the model outputs
00:02:23 something that's not in a reference,
00:02:25 then is scored as incorrect. It's not
00:02:28 nuanced enough to try to judge how
00:02:30 incorrect this word is. For example, if
00:02:32 the word here is replaced with elephant,
00:02:35 then the sentence would make a lot less
00:02:36 sense, but you get the same blue score.
00:02:39 More recently, it's been popular to use
00:02:41 something smarter like a language model
00:02:43 to score references. An example of this
00:02:45 is bird score. Birch score uses a
00:02:48 encoder based LM to get embeddings for
00:02:51 all of the words in the candidate and
00:02:52 the reference sentences and then it
00:02:54 scores how similarly the sequences of
00:02:56 embeddings match up with each other.
00:02:58 This is smarter than an algorithm that
00:03:00 only looks at the words at a surface
00:03:02 level. A different approach is training
00:03:04 a model to predict how similarly would a
00:03:06 human rate these two sentences. This
00:03:09 approach is used in the comet and blurt
00:03:11 evaluation metrics. The comet score
00:03:14 encodes both the reference and the
00:03:15 candidate sentence into an embedding and
00:03:18 then trains a regression model to
00:03:19 predict how similarly would a human rate
00:03:22 these two sentences. If we're going to
00:03:24 use models to predict human preferences,
00:03:26 what if we use humans directly? This
00:03:28 brings us to the second category of
00:03:30 approaches to evaluate LRM outputs using
00:03:34 humans as evaluation. We can do this in
00:03:36 several ways, like having people rate
00:03:39 the quality of an output, or an
00:03:41 increasingly popular approach is present
00:03:43 the user with two options and have them
00:03:45 select which one they prefer more. This
00:03:47 generally feels more fun to use as a
00:03:50 user, and more fun means you will
00:03:52 collect more data. An open- source
00:03:54 version of this idea is the LMS's
00:03:56 chatbot arena. Here, the user inputs a
00:03:59 query and the same query is sent to two
00:04:01 random models. Both models generate a
00:04:04 response and then as a user you can pick
00:04:06 which one you like better. And if you're
00:04:08 not sure after one message, you can keep
00:04:10 chatting with the model until you have
00:04:13 an opinion about which one is better.
00:04:15 Aggregated across millions of user
00:04:17 ratings, they get a score for each
00:04:19 model. This is similar to the ELO score
00:04:21 in chess. Whenever a model is selected
00:04:23 as the winner in a pair-wise
00:04:25 competition, it gains a few rating
00:04:27 points and the other model loses a few
00:04:28 rating points. This way we can get a
00:04:30 leaderboard for how the models all stack
00:04:33 up against each other. Speaking of
00:04:35 chess, I made a video where I train
00:04:38 errors to play chess using reinforcement
00:04:40 learning. You can check it out here.
00:04:42 Anyways, what are the pros and cons of
00:04:45 using humans to evaluate? The most
00:04:47 obvious issue is it's expensive, so you
00:04:50 probably only do it a small number of
00:04:52 times across your entire project. The
00:04:54 second issue is humans are not exactly
00:04:57 perfect annotators anyways and they are
00:05:00 easily affected by surface level tricks
00:05:03 like formatting and the length of the
00:05:05 output. For example, I tried the same
00:05:07 prompt on chatbt and claude. And what I
00:05:10 generally find is chatbt's responses
00:05:13 tend to be longer and have more headings
00:05:15 and bold text whereas claude gives a
00:05:19 good answer but there is no emojis.
00:05:21 There is minimal formatting. So when you
00:05:23 put these two side by side, most users
00:05:25 at a quick glance, they're much more
00:05:27 likely to pick the one that's better
00:05:29 formatted. The solution to this problem
00:05:31 is style control. To do this, they fit a
00:05:34 regression model that incorporates
00:05:36 various attributes that might influence
00:05:39 a user's decision to pick one model over
00:05:41 another, such as the length of the
00:05:43 output or any other factors like style
00:05:46 and headings, um, emojis, formatting,
00:05:49 etc. This way you are able to
00:05:51 disentangle the effect of these style
00:05:54 attributes from the desired interaction
00:05:56 which is how good is the actual model.
00:05:58 The result is some models like clot 3.5
00:06:01 perform much better when the effects of
00:06:04 style is factored out whereas some other
00:06:07 models like GBT40 mini perform worse
00:06:09 after controlling for style. This means
00:06:12 that a lot of the time that this model
00:06:14 beats another model in a head-to-head
00:06:16 competition, it can be attributed to it
00:06:19 having a better formatted output even
00:06:21 though its content is only mediocre.
00:06:24 This solves some of the problems of
00:06:26 human evaluation. But the fact still
00:06:28 remains that getting a lot of human
00:06:30 annotators will be difficult and
00:06:32 expensive. But anything that humans can
00:06:34 do, ALMs can do better. This brings us
00:06:37 to the third category of natural
00:06:40 language evaluation metrics. Using an
00:06:42 LRM as a judge. This means taking an
00:06:45 output that was produced by LM or
00:06:47 sometimes multiple outputs that were
00:06:49 produced by different ARMS and then
00:06:52 feeding this to another LM to a score
00:06:55 instead of giving it to a human. There
00:06:57 are different ways you can set this up.
00:06:59 You can give it two outputs and have it
00:07:02 decide which one it likes better or you
00:07:04 can ask it to rate both of these outputs
00:07:07 on a set of predefined criteria. One
00:07:10 thing that helps is asking the judge LRM
00:07:13 to rate the response on multiple factors
00:07:15 like helpfulness, relevance, accuracy,
00:07:18 depth, creativity, etc. This is just
00:07:21 useful for getting a little bit more
00:07:23 insight into why one output is preferred
00:07:25 over another one. In voice for instance,
00:07:28 we are using LMS to correct some grammar
00:07:31 of some text according to a user
00:07:34 specified style. So the two main things
00:07:36 that we look for during evaluation is
00:07:39 style adherence. How much does the LM
00:07:42 write according to the specified style
00:07:44 and secondly content preservation. How
00:07:46 much does the meaning of the original
00:07:48 text get preserved exactly? You can
00:07:51 train your own writing tone by pasting
00:07:53 in a sample of the text you want to
00:07:54 emulate. Then Voice Writer will correct
00:07:57 your text to mimics that style. You can
00:07:59 speak your thoughts and have the ARM
00:08:01 improve your writing in real time. Try
00:08:04 it for free at
00:08:05 voiceer.io. With that out of the way,
00:08:07 how do we validate this method of using
00:08:09 an ARM as a judge? The usual method is
00:08:12 using interan annotator agreement or how
00:08:15 much do two annotators agree with each
00:08:17 other. The MT bench paper found that
00:08:20 humans on average agreed with each other
00:08:22 81% of the time. But GPT4 agreed with
00:08:26 humans on average 85% of the time. So
00:08:29 the bot's correlation with human
00:08:30 judgments is even greater than two
00:08:32 random humans correlation with each
00:08:34 other. This is proof that the LM is a
00:08:37 reasonable judge in this situation.
00:08:39 That's not to say that LM as a judge are
00:08:42 perfect. In fact, they can be subject to
00:08:44 the similar biases as humans. Biases
00:08:47 such as the length bias like giving
00:08:49 higher scores for longer responses or
00:08:52 the position bias which is giving higher
00:08:54 scores for the first response than the
00:08:56 second response or not bias where a
00:09:00 model rates its own response higher than
00:09:02 other models responses and these should
00:09:04 be controlled for in a similar way as
00:09:06 style control in human evaluations.
00:09:09 Another drawback is cost. Using LMS for
00:09:12 evaluation can be expensive, but
00:09:14 nowadays you have a lot of choice in
00:09:16 which ARM to use as a judge with varying
00:09:19 degrees of cost versus quality. And even
00:09:21 the most expensive ARMS are typically
00:09:23 cheaper than humans. It's also worth
00:09:26 noting that evaluating written responses
00:09:28 is not the only place that arms can be
00:09:31 useful as a judge. They are also
00:09:32 extremely useful in evaluating short
00:09:35 answers that are not like subjective but
00:09:37 is difficult to evaluate using some
00:09:40 fixed rules like the omni math benchmark
00:09:43 where the answer is a mathematical
00:09:45 expression but there can be many
00:09:46 equivalent ways of expressing it. So
00:09:49 this benchmark the use of LM to judge
00:09:51 its correctness. Anyways, that's all I
00:09:53 have for this video, and I hope I gave
00:09:55 you some interesting ideas on how to
00:09:57 pick your LM and evaluate your prompts
00:10:00 for writing quality. If you haven't
00:10:03 already, please like this video, hit the
00:10:05 subscribe button, and ring the bell
00:10:07 icon. This will help my channel grow and
identifier: 2560
model: gemini-2.5-flash-preview-04-17| input-price: 0.15 output-price: 3.5 max-context-length: 128_000
host: 194.230.161.210
https://www.youtube.com/watch?v=Yqaon6YHzaUinclude_comments: None
include_timestamps: 1
include_glossary: None
output_language: en
cost: 0.0115922
Okay, here is the abstract and bulleted summary for the provided transcript about Finale software. *Abstract:* This video provides a critical review and historical analysis of the music notation software Finale, following the announcement of its discontinuation in August 2024. Positioned as a case study on the "death of creative software," the video traces Finale's history from its pioneering 1988 debut, which set industry standards for digital notation (including real-time input and MIDI support), through its long rivalry with Sibelius. Despite its historical importance and eventual flexibility for power users, the video highlights Finale's core problems: a notoriously difficult and unintuitive user interface described as a "usability shanty town," stemming from inconsistent workflows for basic tasks like font changes, note entry, and adding musical markings, compounded by an ancient codebase. The analysis explains Finale's decline through strategic shifts at its parent company (MakeMusic, later Peaksware), including prioritizing the educational platform SmartMusic (and its successor MakeMusicCloud), the adoption of MusicXML as a central format, and a failure to invest sufficiently in modernizing Finale's user experience, ultimately leading to its demise despite a loyal user base. *The Rise and Fall of Finale: A Case Study in Creative Software's Demise* * *0:31 Introduction to Finale's End:* The video focuses on Finale, a music notation software discontinued in August 2024, as a case study on how creative software dies, specifically due to failure to address long-standing usability issues and evolving market dynamics. * *6:49 Pioneering Origins:* Finale was created by Coda Music in 1988 by Phil Farrand and John Borowicz, causing a sensation at NAMM, particularly for its real-time MIDI keyboard input feature. * *7:43 Early Success and Innovation:* Despite a high price (around £3,170 adjusted), Finale became a fully-rounded notation app, defining core features still used today (compose, playback, professional publishing) and supporting the new MIDI protocol. * *9:03 Early User Interface:* The initial UI was described as having a "clean simplicity" inspired by contemporary Mac apps, including tools like the "mass mover" for copying phrases. * *9:50 Initial Difficulty:* From the start, Finale was acknowledged as not being the most straightforward application, often requiring users to spend significant time learning it, utilizing dialog boxes and tutorials. * *12:00 Rivalry with Sibelius:* Finale competed directly with Sibelius (emerging around 1990) and Igor Engraver. Sibelius gained a reputation for being more intuitive, eventually surpassing Finale in ease of use. * *13:04 Declining Development Pace:* Over the 2000s, Finale's development slowed, and its interface felt increasingly outdated compared to modern software. * *13:40 User Experience Critique (Version 27):* The video critiques the modern version's interface as "malnourished," with a "jaundiced colour" and "joyless grey background," arguing its dated appearance contributed to its decline. * *17:23 Frustrating Font Changes:* Changing fonts is highlighted as an incredibly difficult and frustrating process, requiring navigating nested dialogs, holding modifier keys, and encountering limitations like changes only applying to future items. * *19:36 Instrument Name Font Issues:* Changing fonts for instrument names is particularly complex, requiring finding hidden anchor boxes, navigating multiple dialogs, and in some cases, deleting and re-adding the instrument entirely due to missing options (21:08). * *22:09 Difficult Management Dialogs:* Accessing the ScoreManager for adding/deleting instruments involves a dense, unintuitive dialog, and even basic tasks like moving instruments require discovering non-obvious drag handles (23:04). * *24:12 Tempo Marking Complexity:* Editing tempo markings also involves navigating confusing dialogs and understanding specialist jargon (25:04). * *26:23 Power User Bias Critique:* The narrator argues that creative software, including notation apps, often caters too much to power users who rely heavily on shortcuts, neglecting novice and intermediate users who primarily use the UI (28:36). Research is cited showing most users only use a few shortcuts (28:48). * *32:02 Painful Note Entry (Simple Entry):* Finale's "Simple Entry" mode is criticized as highly inflexible and unintuitive, preventing placing notes freely, requiring mode switching to see rests, and behaving like text editing where deleting notes shifts subsequent notes backward (35:06). * *39:04 Speedy Entry Mode:* "Speedy Entry" mode is found to be more straightforward, especially with a MIDI keyboard, though it can still result in notes "bunching up" visually until exiting the mode (40:31). A positive feature is the ability to drag-copy notes/sections (40:54). * *41:11 Complex Markings:* Adding standard musical markings (pedal lines, string numbers, dynamics, etc.) involves convoluted, non-intuitive workflows hidden behind obscure menus, modifier keys, and layered dialogs. * *48:26 Dynamics/Hairpin Difficulty:* Drawing hairpins is demonstrated as a particularly unintuitive process, requiring a rare "double-click and hold and drag" interaction pattern because normal click-and-drag is reserved for selection (51:06). * *52:05 User Testing:* Brief user testing with other YouTubers confirms that even seasoned musicians find basic tasks in Finale difficult and non-obvious. * *53:56 Layout and Spacing Challenges:* Finale requires significant manual intervention for layout and spacing, although it offers precise tools (Measure tool, segment width) once the user understands complex shortcuts and workflows (55:27). * *57:11 Finale's Paradox:* While brutally difficult to learn, mastering Finale provides flexibility, making it arguably the most flexible but also the most time-consuming of major notation apps. * *57:44 Reasons for Decline (Organizational & Technical):* The video explores the behind-the-scenes factors: an ancient codebase that was hard to update (58:25), dependence on third-party plugins that broke with updates (59:36), strategic shifts at parent company MakeMusic/Peaksware prioritizing the educational SmartMusic platform (1:01:22). * *1:06:39 Shift to MakeMusicCloud:* The acquisition of Weezic (browser-based, MusicXML) and Alfred (publisher) led to the creation of MakeMusicCloud, which used MusicXML, eliminating Finale's role as a necessary core technology (1:07:19). * *1:07:51 MuseScore's Growth:* MakeMusic's decision to charge for the beginner version Finale Notepad boosted downloads of free competitor MuseScore, steering younger users away from the Finale ecosystem. * *1:08:38 Discontinuation Announcement and Backlash:* The August 2024 announcement initially included shutting down the software's verification system, causing outrage among users and publishers with Finale catalogues, leading to a quick reversal to allow indefinite authorization (1:09:07). * *1:18:12 Common Bugs and Quirks:* The video concludes with a montage of common, frustrating bugs and inconsistencies experienced by Finale users, highlighting the unrepaired state of the software. I used gemini-2.5-flash-preview-04-17| input-price: 0.15 output-price: 3.5 max-context-length: 128_000 on rocketrecap dot com to summarize the transcript. Cost (if I didn't use the free tier): $0.0116 Input tokens: 38968 Output tokens: 1642
Okay, here is the abstract and bulleted summary for the provided transcript about Finale software.
Abstract:
This video provides a critical review and historical analysis of the music notation software Finale, following the announcement of its discontinuation in August 2024. Positioned as a case study on the "death of creative software," the video traces Finale's history from its pioneering 1988 debut, which set industry standards for digital notation (including real-time input and MIDI support), through its long rivalry with Sibelius. Despite its historical importance and eventual flexibility for power users, the video highlights Finale's core problems: a notoriously difficult and unintuitive user interface described as a "usability shanty town," stemming from inconsistent workflows for basic tasks like font changes, note entry, and adding musical markings, compounded by an ancient codebase. The analysis explains Finale's decline through strategic shifts at its parent company (MakeMusic, later Peaksware), including prioritizing the educational platform SmartMusic (and its successor MakeMusicCloud), the adoption of MusicXML as a central format, and a failure to invest sufficiently in modernizing Finale's user experience, ultimately leading to its demise despite a loyal user base.
The Rise and Fall of Finale: A Case Study in Creative Software's Demise
Below, I will provide input for an example video (comprising of title, description, and transcript, in this order) and the corresponding abstract and summary I expect. Afterward, I will provide a new transcript that I want you to summarize in the same format.
**Please give an abstract of the transcript and then summarize the transcript in a self-contained bullet list format.** Include starting timestamps, important details and key takeaways.
Example Input:
Fluidigm Polaris Part 2- illuminator and camera
mikeselectricstuff
131K subscribers
Subscribed
369
Share
Download
Clip
Save
5,857 views Aug 26, 2024
Fluidigm Polaris part 1 : • Fluidigm Polaris (Part 1) - Biotech g...
Ebay listings: https://www.ebay.co.uk/usr/mikeselect...
Merch https://mikeselectricstuff.creator-sp...
Transcript
Follow along using the transcript.
Show transcript
mikeselectricstuff
131K subscribers
Videos
About
Support on Patreon
40 Comments
@robertwatsonbath
6 hours ago
Thanks Mike. Ooof! - with the level of bodgery going on around 15:48 I think shame would have made me do a board re spin, out of my own pocket if I had to.
1
Reply
@Muonium1
9 hours ago
The green LED looks different from the others and uses phosphor conversion because of the "green gap" problem where green InGaN emitters suffer efficiency droop at high currents. Phosphide based emitters don't start becoming efficient until around 600nm so also can't be used for high power green emitters. See the paper and plot by Matthias Auf der Maur in his 2015 paper on alloy fluctuations in InGaN as the cause of reduced external quantum efficiency at longer (green) wavelengths.
4
Reply
1 reply
@tafsirnahian669
10 hours ago (edited)
Can this be used as an astrophotography camera?
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
6 hours ago
Yes, but may need a shutter to avoid light during readout
Reply
@2010craggy
11 hours ago
Narrowband filters we use in Astronomy (Astrophotography) are sided- they work best passing light in one direction so I guess the arrows on the filter frames indicate which way round to install them in the filter wheel.
1
Reply
@vitukz
12 hours ago
A mate with Channel @extractions&ire could use it
2
Reply
@RobertGallop
19 hours ago
That LED module says it can go up to 28 amps!!! 21 amps for 100%. You should see what it does at 20 amps!
Reply
@Prophes0r
19 hours ago
I had an "Oh SHIT!" moment when I realized that the weird trapezoidal shape of that light guide was for keystone correction of the light source.
Very clever.
6
Reply
@OneBiOzZ
20 hours ago
given the cost of the CCD you think they could have run another PCB for it
9
Reply
@tekvax01
21 hours ago
$20 thousand dollars per minute of run time!
1
Reply
@tekvax01
22 hours ago
"We spared no expense!" John Hammond Jurassic Park.
*(that's why this thing costs the same as a 50-seat Greyhound Bus coach!)
Reply
@florianf4257
22 hours ago
The smearing on the image could be due to the fact that you don't use a shutter, so you see brighter stripes under bright areas of the image as you still iluminate these pixels while the sensor data ist shifted out towards the top. I experienced this effect back at university with a LN-Cooled CCD for Spectroscopy. The stripes disapeared as soon as you used the shutter instead of disabling it in the open position (but fokussing at 100ms integration time and continuous readout with a focal plane shutter isn't much fun).
12
Reply
mikeselectricstuff
·
1 reply
@mikeselectricstuff
12 hours ago
I didn't think of that, but makes sense
2
Reply
@douro20
22 hours ago (edited)
The red LED reminds me of one from Roithner Lasertechnik. I have a Symbol 2D scanner which uses two very bright LEDs from that company, one red and one red-orange. The red-orange is behind a lens which focuses it into an extremely narrow beam.
1
Reply
@RicoElectrico
23 hours ago
PFG is Pulse Flush Gate according to the datasheet.
Reply
@dcallan812
23 hours ago
Very interesting. 2x
Reply
@littleboot_
1 day ago
Cool interesting device
Reply
@dav1dbone
1 day ago
I've stripped large projectors, looks similar, wonder if some of those castings are a magnesium alloy?
Reply
@kevywevvy8833
1 day ago
ironic that some of those Phlatlight modules are used in some of the cheapest disco lights.
1
Reply
1 reply
@bill6255
1 day ago
Great vid - gets right into subject in title, its packed with information, wraps up quickly. Should get a YT award! imho
3
Reply
@JAKOB1977
1 day ago (edited)
The whole sensor module incl. a 5 grand 50mpix sensor for 49 £.. highest bid atm
Though also a limited CCD sensor, but for the right buyer its a steal at these relative low sums.
Architecture Full Frame CCD (Square Pixels)
Total Number of Pixels 8304 (H) × 6220 (V) = 51.6 Mp
Number of Effective Pixels 8208 (H) × 6164 (V) = 50.5 Mp
Number of Active Pixels 8176 (H) × 6132 (V) = 50.1 Mp
Pixel Size 6.0 m (H) × 6.0 m (V)
Active Image Size 49.1 mm (H) × 36.8 mm (V)
61.3 mm (Diagonal),
645 1.1x Optical Format
Aspect Ratio 4:3
Horizontal Outputs 4
Saturation Signal 40.3 ke−
Output Sensitivity 31 V/e−
Quantum Efficiency
KAF−50100−CAA
KAF−50100−AAA
KAF−50100−ABA (with Lens)
22%, 22%, 16% (Peak R, G, B)
25%
62%
Read Noise (f = 18 MHz) 12.5 e−
Dark Signal (T = 60°C) 42 pA/cm2
Dark Current Doubling Temperature 5.7°C
Dynamic Range (f = 18 MHz) 70.2 dB
Estimated Linear Dynamic Range
(f = 18 MHz)
69.3 dB
Charge Transfer Efficiency
Horizontal
Vertical
0.999995
0.999999
Blooming Protection
(4 ms Exposure Time)
800X Saturation Exposure
Maximum Date Rate 18 MHz
Package Ceramic PGA
Cover Glass MAR Coated, 2 Sides or
Clear Glass
Features
• TRUESENSE Transparent Gate Electrode
for High Sensitivity
• Ultra-High Resolution
• Board Dynamic Range
• Low Noise Architecture
• Large Active Imaging Area
Applications
• Digitization
• Mapping/Aerial
• Photography
• Scientific
Thx for the tear down Mike, always a joy
Reply
@martinalooksatthings
1 day ago
15:49 that is some great bodging on of caps, they really didn't want to respin that PCB huh
8
Reply
@RhythmGamer
1 day ago
Was depressed today and then a new mike video dropped and now I’m genuinely happy to get my tear down fix
1
Reply
@dine9093
1 day ago (edited)
Did you transfrom into Mr Blobby for a moment there?
2
Reply
@NickNorton
1 day ago
Thanks Mike. Your videos are always interesting.
5
Reply
@KeritechElectronics
1 day ago
Heavy optics indeed... Spare no expense, cost no object. Splendid build quality. The CCD is a thing of beauty!
1
Reply
@YSoreil
1 day ago
The pricing on that sensor is about right, I looked in to these many years ago when they were still in production since it's the only large sensor you could actually buy. Really cool to see one in the wild.
2
Reply
@snik2pl
1 day ago
That leds look like from led projector
Reply
@vincei4252
1 day ago
TDI = Time Domain Integration ?
1
Reply
@wolpumba4099
1 day ago (edited)
Maybe the camera should not be illuminated during readout.
From the datasheet of the sensor (Onsemi): saturation 40300 electrons, read noise 12.5 electrons per pixel @ 18MHz (quite bad). quantum efficiency 62% (if it has micro lenses), frame rate 1 Hz. lateral overflow drain to prevent blooming protects against 800x (factor increases linearly with exposure time) saturation exposure (32e6 electrons per pixel at 4ms exposure time), microlens has +/- 20 degree acceptance angle
i guess it would be good for astrophotography
4
Reply
@txm100
1 day ago (edited)
Babe wake up a new mikeselectricstuff has dropped!
9
Reply
@vincei4252
1 day ago
That looks like a finger-lakes filter wheel, however, for astronomy they'd never use such a large stepper.
1
Reply
@MRooodddvvv
1 day ago
yaaaaay ! more overcomplicated optical stuff !
4
Reply
1 reply
@NoPegs
1 day ago
He lives!
11
Reply
1 reply
Transcript
0:00
so I've stripped all the bits of the
0:01
optical system so basically we've got
0:03
the uh the camera
0:05
itself which is mounted on this uh very
0:09
complex
0:10
adjustment thing which obviously to set
0:13
you the various tilt and uh alignment
0:15
stuff then there's two of these massive
0:18
lenses I've taken one of these apart I
0:20
think there's something like about eight
0:22
or nine Optical elements in here these
0:25
don't seem to do a great deal in terms
0:26
of electr magnification they're obiously
0:28
just about getting the image to where it
0:29
uh where it needs to be just so that
0:33
goes like that then this Optical block I
0:36
originally thought this was made of some
0:37
s crazy heavy material but it's just
0:39
really the sum of all these Optical bits
0:41
are just ridiculously heavy those lenses
0:43
are about 4 kilos each and then there's
0:45
this very heavy very solid um piece that
0:47
goes in the middle and this is so this
0:49
is the filter wheel assembly with a
0:51
hilariously oversized steper
0:53
motor driving this wheel with these very
0:57
large narrow band filters so we've got
1:00
various different shades of uh
1:03
filters there five Al together that
1:06
one's actually just showing up a silver
1:07
that's actually a a red but fairly low
1:10
transmission orangey red blue green
1:15
there's an excess cover on this side so
1:16
the filters can be accessed and changed
1:19
without taking anything else apart even
1:21
this is like ridiculous it's like solid
1:23
aluminium this is just basically a cover
1:25
the actual wavelengths of these are um
1:27
488 525 570 630 and 700 NM not sure what
1:32
the suffix on that perhaps that's the uh
1:34
the width of the spectral line say these
1:37
are very narrow band filters most of
1:39
them are you very little light through
1:41
so it's still very tight narrow band to
1:43
match the um fluoresence of the dies
1:45
they're using in the biochemical process
1:48
and obviously to reject the light that's
1:49
being fired at it from that Illuminator
1:51
box and then there's a there's a second
1:53
one of these lenses then the actual sort
1:55
of samples below that so uh very serious
1:58
amount of very uh chunky heavy Optics
2:01
okay let's take a look at this light
2:02
source made by company Lumen Dynamics
2:04
who are now part of
2:06
excelitas self-contained unit power
2:08
connector USB and this which one of the
2:11
Cable Bundle said was a TTL interface
2:14
USB wasn't used in uh the fluid
2:17
application output here and I think this
2:19
is an input for um light feedback I
2:21
don't if it's regulated or just a measur
2:23
measurement facility and the uh fiber
2:27
assembly
2:29
Square Inlet there and then there's two
2:32
outputs which have uh lens assemblies
2:35
and this small one which goes back into
2:37
that small Port just Loops out of here
2:40
straight back in So on this side we've
2:42
got the electronics which look pretty
2:44
straightforward we've got a bit of power
2:45
supply stuff over here and we've got
2:48
separate drivers for each wavelength now
2:50
interesting this is clearly been very
2:52
specifically made for this application
2:54
you I was half expecting like say some
2:56
generic drivers that could be used for a
2:58
number of different things but actually
3:00
literally specified the exact wavelength
3:02
on the PCB there is provision here for
3:04
385 NM which isn't populated but this is
3:07
clearly been designed very specifically
3:09
so these four drivers look the same but
3:10
then there's two higher power ones for
3:12
575 and
3:14
520 a slightly bigger heat sink on this
3:16
575 section there a p 24 which is
3:20
providing USB interface USB isolator the
3:23
USB interface just presents as a comport
3:26
I did have a quick look but I didn't
3:27
actually get anything sensible um I did
3:29
dump the Pi code out and there's a few
3:31
you a few sort of commands that you
3:32
could see in text but I didn't actually
3:34
manage to get it working properly I
3:36
found some software for related version
3:38
but it didn't seem to want to talk to it
3:39
but um I say that wasn't used for the
3:41
original application it might be quite
3:42
interesting to get try and get the Run
3:44
hours count out of it and the TTL
3:46
interface looks fairly straightforward
3:48
we've got positions for six opto
3:50
isolators but only five five are
3:52
installed so that corresponds with the
3:54
unused thing so I think this hopefully
3:56
should be as simple as just providing a
3:57
ttrl signal for each color to uh enable
4:00
it a big heat sink here which is there I
4:03
think there's like a big S of metal
4:04
plate through the middle of this that
4:05
all the leads are mounted on the other
4:07
side so this is heat sinking it with a
4:09
air flow from a uh just a fan in here
4:13
obviously don't have the air flow
4:14
anywhere near the Optics so conduction
4:17
cool through to this plate that's then
4:18
uh air cooled got some pots which are
4:21
presumably power
4:22
adjustments okay let's take a look at
4:24
the other side which is uh much more
4:27
interesting see we've got some uh very
4:31
uh neatly Twisted cable assemblies there
4:35
a bunch of leads so we've got one here
4:37
475 up here 430 NM 630 575 and 520
4:44
filters and dcro mirrors a quick way to
4:48
see what's white is if we just shine
4:49
some white light through
4:51
here not sure how it is is to see on the
4:54
camera but shining white light we do
4:55
actually get a bit of red a bit of blue
4:57
some yellow here so the obstacle path
5:00
575 it goes sort of here bounces off
5:03
this mirror and goes out the 520 goes
5:07
sort of down here across here and up
5:09
there 630 goes basically straight
5:13
through
5:15
430 goes across there down there along
5:17
there and the 475 goes down here and
5:20
left this is the light sensing thing
5:22
think here there's just a um I think
5:24
there a photo diode or other sensor
5:26
haven't actually taken that off and
5:28
everything's fixed down to this chunk of
5:31
aluminium which acts as the heat
5:32
spreader that then conducts the heat to
5:33
the back side for the heat
5:35
sink and the actual lead packages all
5:38
look fairly similar except for this one
5:41
on the 575 which looks quite a bit more
5:44
substantial big spay
5:46
Terminals and the interface for this
5:48
turned out to be extremely simple it's
5:50
literally a 5V TTL level to enable each
5:54
color doesn't seem to be any tensity
5:56
control but there are some additional
5:58
pins on that connector that weren't used
5:59
in the through time thing so maybe
6:01
there's some extra lines that control
6:02
that I couldn't find any data on this uh
6:05
unit and the um their current product
6:07
range is quite significantly different
6:09
so we've got the uh blue these
6:13
might may well be saturating the camera
6:16
so they might look a bit weird so that's
6:17
the 430
6:18
blue the 575
6:24
yellow uh
6:26
475 light blue
6:29
the uh 520
6:31
green and the uh 630 red now one
6:36
interesting thing I noticed for the
6:39
575 it's actually it's actually using a
6:42
white lead and then filtering it rather
6:44
than using all the other ones are using
6:46
leads which are the fundamental colors
6:47
but uh this is actually doing white and
6:50
it's a combination of this filter and
6:52
the dichroic mirrors that are turning to
6:55
Yellow if we take the filter out and a
6:57
lot of the a lot of the um blue content
7:00
is going this way the red is going
7:02
straight through these two mirrors so
7:05
this is clearly not reflecting much of
7:08
that so we end up with the yellow coming
7:10
out of uh out of there which is a fairly
7:14
light yellow color which you don't
7:16
really see from high intensity leads so
7:19
that's clearly why they've used the
7:20
white to uh do this power consumption of
7:23
the white is pretty high so going up to
7:25
about 2 and 1 half amps on that color
7:27
whereas most of the other colors are
7:28
only drawing half an amp or so at 24
7:30
volts the uh the green is up to about
7:32
1.2 but say this thing is uh much
7:35
brighter and if you actually run all the
7:38
colors at the same time you get a fairly
7:41
reasonable um looking white coming out
7:43
of it and one thing you might just be
7:45
out to notice is there is some sort
7:46
color banding around here that's not
7:49
getting uh everything s completely
7:51
concentric and I think that's where this
7:53
fiber optic thing comes
7:58
in I'll
8:00
get a couple of Fairly accurately shaped
8:04
very sort of uniform color and looking
8:06
at What's um inside here we've basically
8:09
just got this Square Rod so this is
8:12
clearly yeah the lights just bouncing
8:13
off all the all the various sides to um
8:16
get a nice uniform illumination uh this
8:19
back bit looks like it's all potted so
8:21
nothing I really do to get in there I
8:24
think this is fiber so I have come
8:26
across um cables like this which are
8:27
liquid fill but just looking through the
8:30
end of this it's probably a bit hard to
8:31
see it does look like there fiber ends
8:34
going going on there and so there's this
8:36
feedback thing which is just obviously
8:39
compensating for the any light losses
8:41
through here to get an accurate
8:43
representation of uh the light that's
8:45
been launched out of these two
8:47
fibers and you see uh
8:49
these have got this sort of trapezium
8:54
shape light guides again it's like a
8:56
sort of acrylic or glass light guide
9:00
guess projected just to make the right
9:03
rectangular
9:04
shape and look at this Center assembly
9:07
um the light output doesn't uh change
9:10
whether you feed this in or not so it's
9:11
clear not doing any internal Clos Loop
9:14
control obviously there may well be some
9:16
facility for it to do that but it's not
9:17
being used in this
9:19
application and so this output just
9:21
produces a voltage on the uh outle
9:24
connector proportional to the amount of
9:26
light that's present so there's a little
9:28
diffuser in the back there
9:30
and then there's just some kind of uh
9:33
Optical sensor looks like a
9:35
chip looking at the lead it's a very
9:37
small package on the PCB with this lens
9:40
assembly over the top and these look
9:43
like they're actually on a copper
9:44
Metalized PCB for maximum thermal
9:47
performance and yeah it's a very small
9:49
package looks like it's a ceramic
9:51
package and there's a thermister there
9:53
for temperature monitoring this is the
9:56
475 blue one this is the 520 need to
9:59
Green which is uh rather different OB
10:02
it's a much bigger D with lots of bond
10:04
wise but also this looks like it's using
10:05
a phosphor if I shine a blue light at it
10:08
lights up green so this is actually a
10:10
phosphor conversion green lead which
10:12
I've I've come across before they want
10:15
that specific wavelength so they may be
10:17
easier to tune a phosphor than tune the
10:20
um semiconductor material to get the uh
10:23
right right wavelength from the lead
10:24
directly uh red 630 similar size to the
10:28
blue one or does seem to have a uh a
10:31
lens on top of it there is a sort of red
10:33
coloring to
10:35
the die but that doesn't appear to be
10:38
fluorescent as far as I can
10:39
tell and the white one again a little
10:41
bit different sort of much higher
10:43
current
10:46
connectors a makeer name on that
10:48
connector flot light not sure if that's
10:52
the connector or the lead
10:54
itself and obviously with the phosphor
10:56
and I'd imagine that phosphor may well
10:58
be tuned to get the maximum to the uh 5
11:01
cenm and actually this white one looks
11:04
like a St fairly standard product I just
11:06
found it in Mouse made by luminous
11:09
devices in fact actually I think all
11:11
these are based on various luminous
11:13
devices modules and they're you take
11:17
looks like they taking the nearest
11:18
wavelength and then just using these
11:19
filters to clean it up to get a precise
11:22
uh spectral line out of it so quite a
11:25
nice neat and um extreme
11:30
bright light source uh sure I've got any
11:33
particular use for it so I think this
11:35
might end up on
11:36
eBay but uh very pretty to look out and
11:40
without the uh risk of burning your eyes
11:43
out like you do with lasers so I thought
11:45
it would be interesting to try and
11:46
figure out the runtime of this things
11:48
like this we usually keep some sort
11:49
record of runtime cuz leads degrade over
11:51
time I couldn't get any software to work
11:52
through the USB face but then had a
11:54
thought probably going to be writing the
11:55
runtime periodically to the e s prom so
11:58
I just just scope up that and noticed it
12:00
was doing right every 5 minutes so I
12:02
just ran it for a while periodically
12:04
reading the E squ I just held the pick
12:05
in in reset and um put clip over to read
12:07
the square prom and found it was writing
12:10
one location per color every 5 minutes
12:12
so if one color was on it would write
12:14
that location every 5 minutes and just
12:16
increment it by one so after doing a few
12:18
tests with different colors of different
12:19
time periods it looked extremely
12:21
straightforward it's like a four bite
12:22
count for each color looking at the
12:24
original data that was in it all the
12:26
colors apart from Green were reading
12:28
zero and the green was reading four
12:30
indicating a total 20 minutes run time
12:32
ever if it was turned on run for a short
12:34
time then turned off that might not have
12:36
been counted but even so indicates this
12:37
thing wasn't used a great deal the whole
12:40
s process of doing a run can be several
12:42
hours but it'll only be doing probably
12:43
the Imaging at the end of that so you
12:46
wouldn't expect to be running for a long
12:47
time but say a single color for 20
12:50
minutes over its whole lifetime does
12:52
seem a little bit on the low side okay
12:55
let's look at the camera un fortunately
12:57
I managed to not record any sound when I
12:58
did this it's also a couple of months
13:00
ago so there's going to be a few details
13:02
that I've forgotten so I'm just going to
13:04
dub this over the original footage so um
13:07
take the lid off see this massive great
13:10
heat sink so this is a pel cool camera
13:12
we've got this blower fan producing a
13:14
fair amount of air flow through
13:16
it the connector here there's the ccds
13:19
mounted on the board on the
13:24
right this unplugs so we've got a bit of
13:27
power supply stuff on here
13:29
USB interface I think that's the Cyprus
13:32
microcontroller High speeded USB
13:34
interface there's a zyink spon fpga some
13:40
RAM and there's a couple of ATD
13:42
converters can't quite read what those
13:45
those are but anal
13:47
devices um little bit of bodgery around
13:51
here extra decoupling obviously they
13:53
have having some noise issues this is
13:55
around the ram chip quite a lot of extra
13:57
capacitors been added there
13:59
uh there's a couple of amplifiers prior
14:01
to the HD converter buffers or Andor
14:05
amplifiers taking the CCD
14:08
signal um bit more power spy stuff here
14:11
this is probably all to do with
14:12
generating the various CCD bias voltages
14:14
they uh need quite a lot of exotic
14:18
voltages next board down is just a
14:20
shield and an interconnect
14:24
boardly shielding the power supply stuff
14:26
from some the more sensitive an log
14:28
stuff
14:31
and this is the bottom board which is
14:32
just all power supply
14:34
stuff as you can see tons of capacitors
14:37
or Transformer in
14:42
there and this is the CCD which is a uh
14:47
very impressive thing this is a kf50 100
14:50
originally by true sense then codec
14:53
there ON
14:54
Semiconductor it's 50 megapixels uh the
14:58
only price I could find was this one
15:00
5,000 bucks and the architecture you can
15:03
see there actually two separate halves
15:04
which explains the Dual AZ converters
15:06
and two amplifiers it's literally split
15:08
down the middle and duplicated so it's
15:10
outputting two streams in parallel just
15:13
to keep the bandwidth sensible and it's
15:15
got this amazing um diffraction effects
15:18
it's got micro lenses over the pixel so
15:20
there's there's a bit more Optics going
15:22
on than on a normal
15:25
sensor few more bodges on the CCD board
15:28
including this wire which isn't really
15:29
tacked down very well which is a bit uh
15:32
bit of a mess quite a few bits around
15:34
this board where they've uh tacked
15:36
various bits on which is not super
15:38
impressive looks like CCD drivers on the
15:40
left with those 3 ohm um damping
15:43
resistors on the
15:47
output get a few more little bodges
15:50
around here some of
15:52
the and there's this separator the
15:54
silica gel to keep the moisture down but
15:56
there's this separator that actually
15:58
appears to be cut from piece of
15:59
antistatic
16:04
bag and this sort of thermal block on
16:06
top of this stack of three pel Cola
16:12
modules so as with any Stacks they get
16:16
um larger as they go back towards the
16:18
heat sink because each P's got to not
16:20
only take the heat from the previous but
16:21
also the waste heat which is quite
16:27
significant you see a little temperature
16:29
sensor here that copper block which
16:32
makes contact with the back of the
16:37
CCD and this's the back of the
16:40
pelas this then contacts the heat sink
16:44
on the uh rear there a few thermal pads
16:46
as well for some of the other power
16:47
components on this
16:51
PCB okay I've connected this uh camera
16:54
up I found some drivers on the disc that
16:56
seem to work under Windows 7 couldn't
16:58
get to install under Windows 11 though
17:01
um in the absence of any sort of lens or
17:03
being bothered to the proper amount I've
17:04
just put some f over it and put a little
17:06
pin in there to make a pinhole lens and
17:08
software gives a few options I'm not
17:11
entirely sure what all these are there's
17:12
obviously a clock frequency 22 MHz low
17:15
gain and with PFG no idea what that is
17:19
something something game programmable
17:20
Something game perhaps ver exposure
17:23
types I think focus is just like a
17:25
continuous grab until you tell it to
17:27
stop not entirely sure all these options
17:30
are obviously exposure time uh triggers
17:33
there ex external hardware trigger inut
17:35
you just trigger using a um thing on
17:37
screen so the resolution is 8176 by
17:40
6132 and you can actually bin those
17:42
where you combine multiple pixels to get
17:46
increased gain at the expense of lower
17:48
resolution down this is a 10sec exposure
17:51
obviously of the pin hole it's very uh
17:53
intensitive so we just stand still now
17:56
downloading it there's the uh exposure
17:59
so when it's
18:01
um there's a little status thing down
18:03
here so that tells you the um exposure
18:07
[Applause]
18:09
time it's this is just it
18:15
downloading um it is quite I'm seeing
18:18
quite a lot like smearing I think that I
18:20
don't know whether that's just due to
18:21
pixels overloading or something else I
18:24
mean yeah it's not it's not um out of
18:26
the question that there's something not
18:27
totally right about this camera
18:28
certainly was bodge wise on there um I
18:31
don't I'd imagine a camera like this
18:32
it's got a fairly narrow range of
18:34
intensities that it's happy with I'm not
18:36
going to spend a great deal of time on
18:38
this if you're interested in this camera
18:40
maybe for astronomy or something and
18:42
happy to sort of take the risk of it may
18:44
not be uh perfect I'll um I think I'll
18:47
stick this on eBay along with the
18:48
Illuminator I'll put a link down in the
18:50
description to the listing take your
18:52
chances to grab a bargain so for example
18:54
here we see this vertical streaking so
18:56
I'm not sure how normal that is this is
18:58
on fairly bright scene looking out the
19:02
window if I cut the exposure time down
19:04
on that it's now 1 second
19:07
exposure again most of the image
19:09
disappears again this is looks like it's
19:11
possibly over still overloading here go
19:14
that go down to say say quarter a
19:16
second so again I think there might be
19:19
some Auto gain control going on here um
19:21
this is with the PFG option let's try
19:23
turning that off and see what
19:25
happens so I'm not sure this is actually
19:27
more streaking or which just it's
19:29
cranked up the gain all the dis display
19:31
gray scale to show what um you know the
19:33
range of things that it's captured
19:36
there's one of one of 12 things in the
19:38
software there's um you can see of you
19:40
can't seem to read out the temperature
19:42
of the pelta cooler but you can set the
19:44
temperature and if you said it's a
19:46
different temperature you see the power
19:48
consumption jump up running the cooler
19:50
to get the temperature you requested but
19:52
I can't see anything anywhere that tells
19:54
you whether the cool is at the at the
19:56
temperature other than the power
19:57
consumption going down and there's no
19:59
temperature read out
20:03
here and just some yeah this is just
20:05
sort of very basic software I'm sure
20:07
there's like an API for more
20:09
sophisticated
20:10
applications but so if you know anything
20:12
more about these cameras please um stick
20:14
in the
20:15
comments um incidentally when I was
20:18
editing I didn't notice there was a bent
20:19
pin on the um CCD but I did fix that
20:22
before doing these tests and also
20:24
reactivated the um silica gel desicant
20:26
cuz I noticed it was uh I was getting
20:28
bit of condensation on the window but um
20:31
yeah so a couple of uh interesting but
20:34
maybe not particularly uh useful pieces
20:37
of Kit except for someone that's got a
20:38
very specific use so um I'll stick a
20:42
I'll stick these on eBay put a link in
20:44
the description and say hopefully
20:45
someone could actually make some uh good
20:47
use of these things
Example Output:
**Abstract:**
This video presents Part 2 of a teardown focusing on the optical components of a Fluidigm Polaris biotechnology instrument, specifically the multi-wavelength illuminator and the high-resolution CCD camera.
The Lumen Dynamics illuminator unit is examined in detail, revealing its construction using multiple high-power LEDs (430nm, 475nm, 520nm, 575nm, 630nm) combined via dichroic mirrors and filters. A square fiber optic rod is used to homogenize the light. A notable finding is the use of a phosphor-converted white LED filtered to achieve the 575nm output. The unit features simple TTL activation for each color, conduction cooling, and internal homogenization optics. Analysis of its EEPROM suggests extremely low operational runtime.
The camera module teardown showcases a 50 Megapixel ON Semiconductor KAF-50100 CCD sensor with micro-lenses, cooled by a multi-stage Peltier stack. The control electronics include an FPGA and a USB interface. Significant post-manufacturing modifications ("bodges") are observed on the camera's circuit boards. Basic functional testing using vendor software and a pinhole lens confirms image capture but reveals prominent vertical streaking artifacts, the cause of which remains uncertain (potential overload, readout artifact, or fault).
**Exploring the Fluidigm Polaris: A Detailed Look at its High-End Optics and Camera System**
* **0:00 High-End Optics:** The system utilizes heavy, high-quality lenses and mirrors for precise imaging, weighing around 4 kilos each.
* **0:49 Narrow Band Filters:** A filter wheel with five narrow band filters (488, 525, 570, 630, and 700 nm) ensures accurate fluorescence detection and rejection of excitation light.
* **2:01 Customizable Illumination:** The Lumen Dynamics light source offers five individually controllable LED wavelengths (430, 475, 520, 575, 630 nm) with varying power outputs. The 575nm yellow LED is uniquely achieved using a white LED with filtering.
* **3:45 TTL Control:** The light source is controlled via a simple TTL interface, enabling easy on/off switching for each LED color.
* **12:55 Sophisticated Camera:** The system includes a 50-megapixel Kodak KAI-50100 CCD camera with a Peltier cooling system for reduced noise.
* **14:54 High-Speed Data Transfer:** The camera features dual analog-to-digital converters to manage the high data throughput of the 50-megapixel sensor, which is effectively two 25-megapixel sensors operating in parallel.
* **18:11 Possible Issues:** The video creator noted some potential issues with the camera, including image smearing.
* **18:11 Limited Dynamic Range:** The camera's sensor has a limited dynamic range, making it potentially challenging to capture scenes with a wide range of brightness levels.
* **11:45 Low Runtime:** Internal data suggests the system has seen minimal usage, with only 20 minutes of recorded runtime for the green LED.
* **20:38 Availability on eBay:** Both the illuminator and camera are expected to be listed for sale on eBay.
Here is the real transcript. Please summarize it:
00:00:05 with new opportunities emerging every day,
00:00:09 are getting more options than ever before.
00:00:14 Eric: Um, I’ve an idea!
00:00:16 Eric: Uh, hello, I’m Eric, I’m a second year
00:00:21 Tantacrul: Well, then, take my hand Eric!
00:00:25 Let’s GOOOOOOOOOOOO!!!!!
00:00:31 Tantacrul: PREPARE YOURSELF FOR A NEW DOMAIN OF CREATIVITY!
00:00:34 Eric: Wow…
00:00:36 Eric: Well, it’s a series of short pieces
00:00:40 late renaissance style of Palestrina,
00:00:45 but for a modern twist, I’m relying solely on
00:00:51 Tantacrul: That’s really interesting.
00:00:54 With IdYeah.ai, you can generate an idea at the touch of a button!
00:00:58 and now you have your verse,
00:01:01 bridge and chorus! [Dull Music]
00:01:09 Tantacrul: You gotta have lyrics! Right?
00:01:12 Masses have lyrics, I think. Step forward GenSpekr-module.
00:01:17 and tick the GenAlpha box for max engagement!
00:01:21 And just like that, we got our track! [Song]
00:01:30 Tantacrul: Now let’s find your audience
00:01:34 you got ten thousand followers!
00:01:38 Tantacrul: Yeah, look at the comments:
00:01:41 5 million views!
00:01:46 Tantacrul: It’s blowin up! OK,
00:01:48 Just one click away with PixelChrist!
00:01:52 [Song]
00:01:58 Eric: Wow! I just,
00:02:03 Tantacrul: Hey hey hey! It’s just been remixed by
00:02:07 they’re all stepping to your jam!
00:02:11 Tantacrul: WILL YOU SHUT UP! WILL YOU
00:02:14 NO ONE GIVES A !*%$ ABOUT PALESTINIA.
00:02:19 ABOUT MULTIPHONICS. SQEEKY !*%$ING
00:02:23 NONSENSE. LOOK WHAT I’M DOING LOOK WHAT I’M DOING FOR YOU YOU !*%$ING MORON!
00:02:27 Eric: OK, I’m sorry. I didn’t mean to be ungrateful.
00:02:32 Tantacrul: Hey, the money’s rollin’ in!
00:02:39 Eric: Uh, uh really?
00:02:43 Tantacrul: Yeah, and you’re projected to double that next month.
00:02:45 It’s all happening. You’ve done it Eric.
00:02:51 …just some deductions. This sample is the property
00:02:55 These chord progressions
00:02:59 Those lyrics are the intellectual property of GenSpekker.mdule.
00:03:03 Eric: Oh.
00:03:06 and subscriptions… and….
00:03:11 and you’re also under a few different
00:03:18 Eric: Oh, how am I gonna afford that?
00:03:22 I spoke to my composition professor and he said that
00:03:27 You can do one, you trendy ponce.
00:03:33 They hit their Series A yet?
00:03:37 Are they that Greek startup that generates culturally non-appropriative
00:03:41 there’s a BIG opportunity to be had.
00:03:45 And… hold on, there’s a bit of dust...
00:03:51 [blows dust off the UI]
00:03:53 Tantacrul: What? What is this?
00:03:55 What do these things do?
00:03:59 Not with all these
00:04:03 It doesn’t even have gen alpha toggle. Why would a teacher
00:04:08 HOW CAN ANYONE PUT UP WITH…
00:04:13 You never wanna hear the end is comin’ near
00:04:19 Even tho they shine a light, you don't put up a fight
00:04:26 You know you ain’t yo prime, you just ran outta time
00:04:33 UX is pretty scary, this yo mortuary
00:04:46 great grandmother, Finale was the hot new
00:04:50 and novice composers, publishers, copyists,
00:04:55 It was once the unquestioned industry standard.
00:05:00 However, over 35 years, its star began to fade.
00:05:05 with the pace of technological change,
00:05:09 gradually becoming more outdated and buggy as time wore on. Consequently, it lost market share to
00:05:15 newer competitors over the 2000’s and 2010’s - ultimately leading to a breakpoint in the
00:05:19 summer of 2024, when they announced that they were ceasing development entirely. A decision that came
00:05:24 as a massive shock to its loyal user base. So, in this video, I want to look at how Finale
00:05:29 got started, what was happening behind the scenes and how it got swept away due
00:05:33 to larger external forces that its creators could not have predicted. And ultimately,
00:05:37 how, when the writing was on the wall - when it was clear that big improvements were needed - it
00:05:41 found itself stuck. Unable to change. This is a case study in how creative software dies.
00:05:47 For those new to this channel, I’m a designer with over a decade of experience leading teams
00:05:51 that build creation software - including a notation app. And using this experience,
00:05:55 us understand what it’s like to use,
00:06:00 which in turn will help us understand how it ultimately fell from grace.
00:06:04 Beyond anything else, this video is a cautionary tale. A tale of what happens
00:06:08 when you fail to address long standing problems. Stacking feature after feature on top of shaky
00:06:13 foundations. Producing what I think of as a usability shanty town. Stitched together with
00:06:18 no rhyme or reason. Streets filled with dead ends, bridges to nowhere and little windows you
00:06:23 things: young children, with tears
00:06:28 in their eyes, kneeling down to light candles and placing flowers next to your grave.
00:06:35 And it’s at that moment, you come to the realisation “this can be avoided.”
00:06:49 In 1988, Phil Farrand and John Borowicz, who worked for the company Coda music,
00:06:54 Farrand, the driving force
00:06:58 of digital notation tools at the
00:07:02 create something better. And since
00:07:07 this was his only goal, having no intention to ever work in software engineering again,
00:07:11 he named the app Finale. When demoed at NAMM in Atlanta
00:07:15 in that same year, it caused a sensation - especially due to its ability to notate
00:07:20 performances on a keyboard in real time. A tremendous number of people seem interested
00:07:24 in the idea of, sort of, this dream transcription machine, where they play a keyboard and, sort of,
00:07:29 in real time, out the other end, pops sheet music.
00:07:32 glowing:
00:07:39 Interestingly, in that same article,
00:07:43 it lists the price in the UK as £995, which is roughly £3,170 in today’s money. Wow!
00:07:51 Finale came at the tail end of a decade which had seen an explosion of notation apps for the
00:07:56 macintosh. The big differentiator for Finale was how it convincingly solved many of the
00:08:00 For example, there had already been a few realtime
00:08:05 the earliest being Mockingbird, released in 1980. And even
00:08:09 the app wasn’t practical for real world use because it was
00:08:14 than simply writing the music out on paper.
00:08:18 Finale, on the other hand, was the first fully-rounded concept for a notation app, defining
00:08:23 the core components of digital sheet music writing that still persist today. You could compose your
00:08:28 music in it using various different methods. You could listen back to the music you’d written to
00:08:32 get a sense of how the piece might sound when performed and you could finesse the layout and
00:08:36 engraving to a high enough standard that it could be professionally published. To sweeten the deal,
00:08:40 it also supported an exciting new protocol called MIDI, which made it possible for composers to
00:08:45 “The most basic use of MIDI is to connect
00:08:49 Play one instrument and the information about what you’re playing
00:08:54 is sent to the other instrument which will precisely mirror your performance”.
00:08:58 expanding its convenience and usefulness to
00:09:03 The initial interface of Finale had a
00:09:06 clearly taking inspiration from contemporary
00:09:10 It also came with all kinds of useful time saving
00:09:16 Composer: Using the mass mover tool. Just drag
00:09:21 And… a copy of that melodic phrase will then
00:09:31 Finale wasn't the first notation app to feature
00:09:35 Similar interfaces had been introduced by
00:09:39 (later renamed Mosaic) and Encore,
00:09:45 Anyway, despite all this competition, Finale
00:09:50 However, it wasn’t without its problems. Right
00:09:53 was not the most straight-forward application to
00:09:58 the user with sequences of complicated dialog boxes.
00:10:11 although to some extent, we can
00:10:16 Ferrand: And it has a number of tools and
00:10:20 There’s going to have to be a lot of time that they need to spend with
00:10:24 Now, because new generations of personal
00:10:28 Ferrand and Borowicz recognised
00:10:32 Finale so it wouldn’t be dependent on any specific
00:10:36 cross-platform engine called ‘ENIGMA’ which was
00:10:41 Borowicz: We determined when we started
00:10:44 Enigma that we weren’t going to be concerned about hardware constraints
00:10:47 at all. We were going to build it the way we thought it should be built.
00:10:50 a version of Finale for Windows two
00:10:54 was shooting up in popularity,
00:10:59 with the massive release of Windows 3.0 just around the corner.
00:11:02 two new notation apps emerged that also made a splash.
00:11:07 The first was Sibelius, initially developed for the Acorn Archimedes computer - a niche machine that
00:11:12 The second was Igor Engraver, an app that successfully found financing
00:11:18 Finale could be toppled by a competitor
00:11:22 Peter Bengtson, the creator of Igor Engraver, called Finale ‘as
00:11:28 Ultimately, it turned out the Igor Engraver team were right…
00:11:32 Over the years, some of the other competitors,
00:11:36 like Mosaic and Encore gradually fell out of usage. Igor Engraver,
00:11:40 was also eventually knocked out of the race when
00:11:44 Incidentally, according to Bengtson,
00:11:49 might be bought by Steinberg in 2001
00:11:54 in part due to the financial turmoil caused by 9/11.
00:12:00 This left Finale and Sibelius. The two titans of the industry, who set about expanding their
00:12:05 for digital notation software.
00:12:09 a free, simplified version designed to
00:12:15 cater to musicians who didn't need the full functionality of the professional
00:12:18 introduce new users to their ecosystem,
00:12:23 go on to buy the full version as their needs matured.
00:12:26 The rivalry between Finale and Sibelius was pretty much the stuff of legend.
00:12:31 and institutions strongly advocating for one or
00:12:36 it: “Today such partisanship is not uncommon in
00:12:42 musicians' circles, and may be expressed in tones associated more with the sports
00:12:46 However, over time, throughout the 2000’s,
00:12:50 which had a deserved reputation for being much
00:12:55 Finale’s rate of development slowed down and
00:13:00 a modern notation app. The reasons for its decline
00:13:05 as well as the company’s inability to turn things around are numerous -
00:13:08 and we’ll be digging into that later. Regardless, it needs to be stated again
00:13:12 The app that set the standard. And this poses a bit
00:13:17 of a problem when trying to assess it today due to a tension that exists between its glory days and
00:13:22 its more recent history, where it gradually became a source of disappointment and even ridicule.
00:13:26 review about design, it's important
00:13:30 This was an app, let’s not forget,
00:13:35 that people were paying for and that many teachers were insisting their students use over other
00:13:39 competitors until as recently as the summer of 2024. So let’s get acquainted with it,
00:13:40 version: Version 27. Now for those new to this channel, as a designer,
00:13:55 I always assess software by looking at it from the perspective of a new user. When an app is
00:14:00 well designed, it’ll be easy for people to pick up, without having to constantly refer
00:14:04 to a manual or search through online forums. And this isn’t merely abstract design ideology.
00:14:10 It’s important from a business perspective too because when operating in a competitive
00:14:14 landscape - and believe me the notation space is seriously competitive - it needs to be highly
00:14:18 intuitive because your potential customers always have the option of going somewhere else.
00:14:22 So let’s start with the basics. Creating a new score. To do this properly, I set myself the goal
00:14:27 of recreating a piece I composed in 2024 for a group called the Spaceship Ensemble, who premiered
00:14:33 it in the Hague as part of the Festival Dag in de Branding. I’m choosing this score because it uses
00:14:45 a diverse range of instruments along with a few moments where the music gets quite dense. However,
00:14:49 it’s nothing too crazy. The notation style is conventional and there are no 20th century
00:14:54 techniques at play. This score should be pretty straightforward to create.
00:14:57 Things started off OK. When you opt to create a new score, Finale provides a simple screen where
00:15:02 you can choose from a bunch of templates, or you can add instruments manually.
00:15:06 adding: marimba, celesta, accordion, electric piano, regular piano,
00:15:11 harp, electric guitar and harpsichord. Next, I arranged the ordering of the instruments using
00:15:15 these two buttons (that’s foreshadowing there. Remember that). Then I filled in the score
00:15:20 details along with the tempo, time signature and such. I didn’t pay enough attention to
00:15:24 what was being asked here, so I set the tempo in two different places, like a gobshite.
00:15:28 And with that done, let’s just stop and look at Finale. Hold on, let me
00:15:31 just get rid of some of this dust. (Blow) Now, let’s start with the most obvious positive.
00:15:36 Finale’s interface is very, very minimal, giving about as much room to the score as
00:15:41 is possible. Now, on the overall appearance side, what can you say? Finale doesn’t just
00:15:46 look dated. It looks malnourished. The score has this weird, jaundiced colour, and combined with
00:15:51 the sickly green selection colour and joyless grey background, it just looks exhausted.
00:15:57 Finale: Oh, uh, are you talking about me… who are you? Oh,
00:16:04 Tantacrul: It’s OK Finale. Nothing’s
00:16:08 going on. Here, have a Werther’s Original and go back to sleep.
00:16:12 Finale: Uh. Thanks. Wake me up when there’s an update. I feel like it’s
00:16:17 Tantacrul: Sure thing Finale. You
00:16:20 just sleep now and it’s all gonna be OK. This piece I’m writing is gonna be a real
00:16:24 Finale: I like bangers… bangers. Bangers
00:16:30 and mash. I remember once… OK, I think… I think it's gone
00:16:39 back to sleep. Great, let's continue making fun of how it looks.
00:16:43 here: something composers, regardless of how technical they are,
00:16:47 on: Finale looks worryingly behind the curve. In my opinion,
00:16:52 especially from speaking to other composers and reading forums about it, this will have
00:16:56 contributed to Finale’s decline because for the last 15 years or more, it has noticeably looked
00:17:00 like an app that is not in active production. Now, before I get started actually writing music,
00:17:05 I gotta sort a few other things out. First, let’s switch the notation font to Leland because that’s
00:17:09 what I’m using in MuseScore. Second, there’s no way I’m going to accept Times New Roman as my
00:17:14 default font. That’s another thing. Come on, guys… Times New Roman? Times is being used in the title,
00:17:19 the composer name, the footer, the performance directions and the instrument names. So let’s
00:17:23 switch that too. And this was the point where things began to go south. Little did
00:17:27 I realise what an incredibly frustrating odyssey I was about to embark on.
00:17:31 OK, are you ready for the pain? Now, the ideal way to change the score
00:17:36 font would be via some kind of global setting that lets me change all instances of Times New Roman
00:17:41 to something else. But where might that be? I searched for a while but eventually lost patience
00:17:46 and had to consult a few tutorials. OK, first, click on Document, then ‘Document options’ and
00:17:51 then ‘Fonts’. From there, next to ‘Text’, you’ll find an option to set the font for certain things,
00:17:56 including the instrument names. You can also change it for lyrics. Then you press this button
00:18:00 to set the font. Great. Although, you have to do this for every item. That’s 11 times in total.
00:18:05 With that done, I then pressed the apply button and… there was no change. Hold on, what? Why
00:18:11 didn’t that work? Well, I had been a silly goose, you see because I’d failed to read this little
00:18:15 dialog: “Will only affect items created after font changes are applied.
00:18:21 Existing items will not be changed’. OK, that is REALLY weird. The app can’t apply font
00:18:26 changes to some things that are already on the score? Worse still, the developers were aware it
00:18:30 was a problem and left us this little semi-apology note, rather than just make it work?
00:18:34 Well, alright, if I can’t apply font changes globally,
00:18:37 let’s figure out how to do it manually. To start, I tried right clicking on the title.
00:18:41 None of these items are what I need. Next, I tried double-clicking on the title in the
00:18:45 hope that it might trigger some kind of font UI - perhaps something that might look like
00:18:49 this. I tried everything I could think of but eventually gave up and went online to
00:18:52 find a relevant tutorial. That’s right. I needed a tutorial to figure out how to change the font.
00:18:58 So, what did I learn? Well, you need to double-click on the text first to
00:19:02 enter into an editing mode, then select the characters you want to change. At this point,
00:19:06 you’d be forgiven for not noticing that a sneaky new ‘Text’ option has suddenly materialised in the
00:19:11 top menu up here. Select that, and you’ll find the options to specify the font, which I hope
00:19:15 we can all agree is completely crazy. Exhibit A on the downsides of having such a sparse UI.
00:19:21 Anyway, once I changed the text, it only affected the title on the first page and not the smaller
00:19:25 titles on every other page, so I needed to manually go and change those too. Same with the
00:19:30 composer name. Next, I changed the footer using the same process. I think I’m getting the hang
00:19:36 of this now. So, what about the instrument names? Well, no such luck. The process for this is just,
00:19:42 frankly, depressing - the first but by no means the last example of how poorly stitched together
00:19:47 this application really is. This took me over 30 minutes to figure out how to solve. So,
00:19:52 here’s what you do. First, double-click on the instrument name to select it. This triggers a
00:19:56 collection of miniature boxes, which look like anchors. In fact, they are anchors, which allow
00:20:01 you to move the names around. But looks can be deceiving because these boxes have a special
00:20:05 hidden power. All you have to do is develop superhuman levels of creative problem solving and
00:20:09 intuit to double-click on this specific box, which brings up a dialog called ‘Group Attributes’.
00:20:15 So, let’s look at the top option up here, “Full Group Name” and you’ll see an edit button to the
00:20:20 right. Click on that and you get another modal dialog, which maddeningly tells you what the
00:20:24 font being used is but doesn’t let you actually change it. So, if you want to change the font, you
00:20:29 have to select the text here in full… then somehow notice that the ‘Text’ option has again sneakily
00:20:34 popped into existence up here in the top menu. From there, you’ll be able to specify your font.
00:20:39 I have literally never seen a dialog interact with the main menu bar before,
00:20:43 so it never occurred to me that this is how it might work. It’s bonkers.
00:20:46 And this leads us to the next problem. Since the text change only applied to the instrument
00:20:51 I selected, and worse, only applied to the instrument name on the first page and not the
00:20:55 abbreviated name on all subsequent pages. You have to change those separately. So,
00:20:59 because I have 8 instruments - each with long and abbreviated names - that means I have to
00:21:03 specify my desired font 16 separate times. Thank heavens I didn’t choose an orchestral piece.
00:21:08 And this is when you enter into a kind of a jank chain reaction, where the solution to one problem
00:21:12 requires you to use other parts of the app which contain multiple new kinds of obstacles. For
00:21:17 example, FOR SOME REASON, I can change the font style for every instrument on this score… apart
00:21:22 from the guitar. This is because the guitar dialog is completely different to all the others and
00:21:26 when putting it together, it seems they forgot to include the option to change the font styling. So,
00:21:31 there appears to be no way globally or locally to change the guitar name away from Times New Roman.
00:21:36 But I’ll be damned if I’m going to let this defeat me, so… I realised that my only remaining
00:21:40 option was to delete this guitar and add a new one instead, which because it’ll be a
00:21:44 newly generated instrument should now inherit that global default I set up earlier.
00:21:49 And this got me into even more trouble because nothing is where you expect it to be. Either
00:21:54 hidden up here somewhere in this miserable menu bar or by clicking on these blank,
00:21:57 miniature loot boxes. Eventually, I figured out that the option I wanted
00:22:01 was up here in ‘Window > ScoreManager’. This brings up an eency and very dense looking
00:22:06 dialog. Now we click on this ‘Add Instrument’ button, which allows me to add a guitar. Ah,
00:22:10 OK, it’s been added in a position I don’t want… so all I have to do now is move it up here. However,
00:22:16 again, another blocker. I couldn’t figure out how to move this guitar up into the desired position.
00:22:20 I tried dragging and got this popup. I looked down here at all this mess and couldn’t find an up or
00:22:25 down button. Particularly strange because, as we saw earlier, the dialog we used to add instruments
00:22:30 in the first place had up and down buttons, which is what led me to expect them here. But you can
00:22:35 only use that first dialog once when setting up your score so I couldn’t access it again.
00:22:39 This is another notable problem with Finale. Layers upon layers of different dialogs from
00:22:43 different generations all partially doing the same thing. This one was made during the first
00:22:47 Gulf war. This one dates to the Arab Spring. Anyway, you wouldn’t believe how much trouble
00:22:51 I had just trying to move this instrument up to the right place. I checked multiple forum threads
00:22:56 and searched on YouTube for ages. Eventually, I saw someone do it in a video tutorial… and… oh,
00:23:01 hold on, it turns out the instruments can actually be dragged… I was just clicking
00:23:04 in the wrong place. What I should have done is click and hold on this symbol over here,
00:23:08 which I had previously interpreted to be a menu icon, which is why I steered clear of it.
00:23:12 Incidentally, this icon is meant to be a grip or what some would call a drag handle. But it
00:23:17 doesn’t really look like something that can be dragged, does it? It looks like some old
00:23:20 iTunes icon or something like that. Right, and all that’s left now is to delete
00:23:25 the original guitar. Oh God… pressing the delete key doesn’t do anything. Right-click doesn’t do
00:23:30 anything. There’s no delete button down here, like… again, it’s there in this dialog. What’s
00:23:34 going on? This one screwed me up even worse than the previous problem. Another frantic web hunt
00:23:39 later and I finally learned how to do it. Guess where the option is? It is actually here. Go on,
00:23:44 have a look now - feel free to pause and you tell me where I go to delete this guitar?
00:23:49 Figured it out? Well, if using a mouse, hold shift and scroll (or sideways scroll on trackpad),
00:23:54 to move everything to the right… at which point a horizontal scrollbar manifests that wasn’t
00:23:59 there before, and it’s revealed that there were actually more options out of view. And
00:24:03 right at the end, there’s this miserable little ‘X’ button, which allows you to
00:24:06 delete the instrument. JESUS CHRIST. Lastly, if you remember, I made a little
00:24:12 mistake during the creation of my score which produced this weird tempo marking you see here,
00:24:16 which says ‘116’ twice and is using Times New Roman besides. So, let’s change that. First,
00:24:21 I double-click on it and it launches this dialog box which contains a bazillion symbols
00:24:26 in it. I don’t want any of these, so I just select ‘Edit’. But that’s no good because I’m
00:24:30 simply editing the first thing selected in this list. OK. So (sigh), right, tutorial time again.
00:24:37 OK, what I should have done… see this eency little anchor box? Yeah, double-click on that,
00:24:42 which brings up the same list we saw before - although now, I can see my guy pre-selected, right
00:24:47 at the bottom of the list. Skill unlocked! Right, now, let’s click on it and… another mad
00:24:53 looking dialog. Let’s just delete the unneeded text. And now let’s change the font. Ah,
00:24:58 checkbox:
00:25:04 ‘Use tempo marks category font’... like.. because I’ve designed notation fonts and
00:25:09 have years of experience working on notation apps… I have a sense of what this means but
00:25:13 how a regular user is meant to understand this specialist jargon is beyond me.
00:25:17 So, let’s uncheck it. Ah, that’s interesting. Now the dialog, for some reason, has a normal
00:25:22 font selection widget in it. So, hold on… they actually built a font selection
00:25:26 UI in Finale… but only use it here? Why not everywhere else and do away with this sneaky,
00:25:30 phantom option that appears in the menu? Now, for experienced Finale users shouting at
00:25:35 their screens, I did find some additional font options for score text buried away under a button
00:25:39 named ‘edit categories’. This was useful because it allowed me to keep things consistent going
00:25:43 forwards. But regardless of the utility of some of these options, I hope I’ve demonstrated how
00:25:48 baffling and disorganised the process of modifying a score really is, with relevant options scattered
00:25:53 to the wind rather than kept in one place. Anyway, with all that said, I had finally managed
00:25:57 to get rid of Times New Roman completely. And I have to admit that I felt a degree of satisfaction
00:26:01 in getting there… finally. I had scaled Mount Kilimanjankro… finally... And what I was left with
00:26:07 achieved in the face of massive
00:26:12 and unexpected adversity.
00:26:14 FINAL(L)E(Y).
00:26:23 about in my other videos that I think
00:26:30 See, in my previous videos, especially the
00:26:35 I spent time assessing the systems for note input from the
00:26:40 not from the point of view of a power user
00:26:43 Power users - unless they’ve had a lot of
00:26:47 experience actually building apps professionally themselves - tend to have a blindspot about this.
00:26:52 They tend to dominate discussions on forums and often don’t realise that they are massively
00:26:56 overrepresented there. Power users are a minority. People who hang around software forums shooting
00:27:00 the breeze about macros and shortcuts. I learned the hard way how you should avoid
00:27:04 thinking like a power user and burying functionality behind shortcuts back when
00:27:07 I was working on Paint 3D - the first creation app I worked on. I’m going to
00:27:11 give a very simple example here. You see, because I was a professional designer,
00:27:15 I had incorporated tons of shortcuts into my own workflow on other apps like Illustrator,
00:27:19 Photoshop and Cinema 4D, not fully appreciating how much of a personal bias this really was. So,
00:27:24 when I began sketching out the overall layout of Paint 3D, I intentionally left out dedicated
00:27:29 undo and redo buttons in order to prevent the app from looking too cluttered. I did this because I
00:27:33 thought that shortcuts like Ctrl+Z had entered into common usage. And I argued, ‘Undo’ can be
00:27:38 found in the Edit menu. My colleagues, many of them also professional designers, agreed.
00:27:42 So, a little later on, once we’d built our first prototype, we held a user testing session with a
00:27:47 group of 5 people. During the session, 3 of those people went looking for undo and redo buttons
00:27:51 in the UI. I was surprised by this but wrote it off because the sample size was so small. But it
00:27:56 kept happening in successive sessions, with such consistency that we eventually decided to include
00:28:00 undo and redo buttons on the top right of the app. Then, when Paint 3D was eventually launched,
00:28:05 we started getting raw data about how it was being used in the wild. And what we discovered was that
00:28:09 not only did the overwhelming majority of our users prefer the physical undo button over Ctrl+Z,
00:28:14 the undo button was actually the most clicked on UI element in the entire application. This
00:28:18 is why I kept these same buttons in MuseScore 4 and why we didn’t remove them when designing
00:28:22 Audacity 4 either. Oh yeah, this is what Audacity 4 looks like by the way. It’s not out yet.
00:28:27 And this is just one of the more easy to explain examples of something I’ve seen
00:28:31 played out soooo many times over a decade of working on desktop creation apps. The tendency
00:28:36 for power users to design applications for other power users, when in fact,
00:28:39 the majority of users tend to stick to between 1-10 shortcuts, which includes standard shortcuts
00:28:44 for saving, copying and pasting, etc. And this isn’t just my opinion. It’s borne
00:28:48 out in research too, some of which I’ve linked to in the description.
00:28:52 One study found that professionals using Microsoft Word overwhelmingly
00:28:55 preferred to use app iconography rather than shortcuts, even though learning shortcuts would
00:28:59 help them work more efficiently.
00:29:11 The study also neatly summarises in one sentence,
00:29:14 design:
00:29:26 And this gets to the crux of my design reviews.
00:29:30 It’s whether or not people can figure
00:29:34 That’s the problem to crack in order to
00:29:38 Shortcuts may be more efficient but they’re also - literally
00:29:43 even go looking for it. This is why - on
00:29:48 no feature should be accessible only via a shortcut.
00:29:52 “yeah, but notation apps are quite complex, so shortcuts are probably
00:29:57 And you wouldn’t be wrong. You do see an increase in
00:30:01 of users still use between 1-10 shortcuts.
00:30:07 of composers over the years is
00:30:10 Averaging the various polls I put out,
00:30:14 The percentage of users who said they use
00:30:18 Those who said they use between 10-20 shortcuts
00:30:23 more than 20 shortcuts were 17%.
00:30:28 perspective too. Many students just starting out
00:30:32 especially if they’re
00:30:35 and will need time to learn to become a power user.
00:30:39 in the long-term to ignore this audience.
00:30:43 you have to cater to power users and you have
00:30:46 to cater to point-and-click users too.
00:30:50 particularly funny when some Dorico users
00:30:55 review a few years ago because I focused on the
00:30:59 some of the things that are possible when you
00:31:03 Some even went as far as to suggest - you will be shocked to
00:31:07 shortcuts so I’m just so dumb!
00:31:12 Anyway, I actually wrote a whole section about shortcuts for that video - you’re watching
00:31:16 because the video was already really long,
00:31:20 app that - as I said at the time, I rather liked.
00:31:25 think “he misrepresented my favourite app”,
00:31:29 I do understand your anger. And that’s OK.
00:31:33 It's OK to be wrong.
00:31:34 Seriously though, congratulations on knowing so many shortcuts.
00:31:37 Actually, while we’re on the topic of Dorico,
00:31:42 my video should not play a big part in your decision.
00:31:46 of good changes since. So if you’re on the market,
00:31:51 go get the free trial and see for yourself… and then watch my video.
00:32:02 Finale: how you enter notation. As before, I’m going to
00:32:08 try working things out for myself, seeking help only when I get really stuck. To begin writing,
00:32:13 there are two modes to choose from. Simple Entry and Speedy Entry. Let’s start with
00:32:17 Simple. And right away, we encounter an awkwardness that’s unique to Finale.
00:32:21 Look at this part of the score I’m attempting to copy. Here,
00:32:23 you can see a short silence at the beginning of the bar, followed by a simple offbeat rhythm.
00:32:27 That’s what I’d like to enter in Finale. Now, for all the other major notation apps, once
00:32:31 you place a note on the score, a bunch of rests are also created. This is useful because rests
00:32:36 act as precise rhythmical markers that help you place other notes more accurately. In Sibelius,
00:32:41 I’ve placed this eighth note a little earlier in the bar than I intended. But because these
00:32:45 rests are here, I can now quickly fix the problem by writing another eighth note here,
00:32:49 which breaks open this rest. Now I can enter my desired rhythm, delete the first two notes,
00:32:54 clean up the first rest, and I’m done. An experienced person would work much more
00:32:58 efficiently than this of course… but for a user just getting started, you’ll need to make these
00:33:02 kinds of corrections all the time and it’s nice when the app is flexible enough to allow you.
00:33:05 In Finale, when you’re using Simple Entry, you’re blocked in so many different ways. First off,
00:33:10 when you place a note on a blank measure, it can only appear on the first beat of the bar,
00:33:14 and once it is placed, no rests appear after it to help guide you. You can use multiple shortcuts
00:33:18 to get around this. But because we’re not digging into shortcuts quite yet - this is Simple Entry
00:33:23 after all - the only thing you can do is add the next note from left to right. If you try
00:33:27 to add a note later in the bar, it won’t do it. You eventually figure out that if you exit Simple
00:33:31 Entry mode by pressing ‘escape’, Finale performs a little calculation, after which, rests actually
00:33:36 do appear in the measure, which is weird. So, if you enter back into Simple Entry mode again,
00:33:41 now you can place your desired notes on the rests. However, beware. If you’re slightly imprecise,
00:33:46 you get this infuriating noise/ Now, I was unfortunate here because
00:33:50 one of the first things I tried to do when I realised I could only enter notes from left
00:33:53 to right was to try replacing this first note with a dotted quarter rest, which is what’s
00:33:57 actually needed here. However, instead of placing a dotted quarter rest, Finale said [NOISE]. And
00:34:05 I really didn’t understand why. I interpreted this to mean ‘you can’t enter rests that way’,
00:34:09 so moved on to try other approaches. We’ll get back to this rest problem in a minute.
00:34:13 So, my next approach was - clutching at straws - to simply add another dummy note here,
00:34:18 which broke open this rest and allowed me to enter my desired rhythm. And once I did that,
00:34:22 my intention was to delete these dummy notes. But even selecting and deleting notes is hard. First,
00:34:26 I exited Simple Entry mode, which takes me back to the selection mode. This
00:34:30 allowed me to select a note, which you can see highlighted in green. Then I pressed delete,
00:34:34 ah, and it doesn’t do anything. If I try holding shift to extend the selection to these notes,
00:34:38 the whole measure turns green, which is again something I don’t want. Uh, this is frustrating.
00:34:43 So then I try double-clicking on a note. OK, I guess I need to clear this green selection first
00:34:48 by clicking away. Cool. Let’s try double-clicking again. OK. But it’s a different kind of selection
00:34:53 but at least, I can finally delete the note. Or at least that’s what I thought because Finale
00:34:58 still found a way to screw me up. Turns out that when you delete a note, it brings all subsequent
00:35:02 notes in the bar backwards to fill its place. Meaning that when I clear away the notes I didn’t
00:35:06 want, my desired rhythm is dragged back out of position. This approach - to treat notation like
00:35:10 text editing - is a relic of the past. It has been decidedly moved away from. When you delete a note,
00:35:16 subsequent notes should stay put. Why? Because words are not notes. If I delete
00:35:20 a word in a sentence, it shouldn’t leave a big horizontal gap because that doesn’t serve
00:35:24 In music, if we delete a note,
00:35:27 legitimate communicative function in music.
00:35:32 and you have a bunch of chords laid out - and you think:
00:35:35 Like, that’s what this is.
00:35:39 Now, had I started all of this by simply adding a rest before anything else,
00:35:42 But let me remind you, there was a point earlier
00:35:46 [NOISE]... was turned away
00:35:50 allowed to enter a rest. All I knew was ‘NO’.
00:36:09 do the most basic musical operations
00:36:13 a highly inflexible and uncommunicative opponent.
00:36:17 writing are just fundamentally not there.
00:36:22 Or at least, not by the standards expected from the mid 2000’s onwards.
00:36:26 agree with this observation. Some will say “you
00:36:30 marquee selected the notes and then press delete”.
00:36:35 but I did use that tool to just select a note and press delete.
00:36:39 Some may think ‘but Finale has its own unique concepts.
00:36:44 That’s true… but its uniqueness is basically that it’s
00:36:48 And there are no points for being weird in this field.
00:36:52 had been adopted as the default decades ago,
00:36:57 That would be what you’d expect. And I can’t stress this enough.
00:37:02 between two apps, say Finale and Sibelius,
00:37:06 a little more permissive and forgiving,
00:37:10 that’s often all you need to make a choice.
00:37:15 after a few days of flopping around, I decided it was
00:37:19 The first I learned, which is absolutely
00:37:23 Samaritan on Twitter, is the ‘R’ shortcut,
00:37:27 That sped things up massively.
00:37:31 moving the music, I discovered a new problem.
00:37:35 It should be the other way around.
00:37:39 note rest. So, I changed the half to
00:37:43 that Finale would do the calculation to fix
00:37:47 So now, I just want to replace these last two
00:37:51 Ugh. So… even though there’s room for a half note rest,
00:37:58 So, to get the result I wanted,
00:38:03 rests and then added the correct rest.
00:38:10 The problem is, in Finale, even
00:38:14 duration, like this quarter note, with a longer
00:38:19 A limitation that turns note entry into a complex sudoku puzzle.
00:38:24 I now have the wrong rests at the beginning of a measure,
00:38:28 As we’ve already established, if I try to replace
00:38:33 However, if I now try to delete these first two rests,
00:38:38 which I was happy with, have shunted back
00:38:43 The only solution - I eventually discovered, was to select
00:38:47 then marquee drag to select
00:38:52 Finale then usually recalculates the measure in such
00:38:57 So, in summary, ‘simple entry’ is a complete mess.
00:39:01 The most unintuitive point & click system I have ever used.
00:39:04 For years, I’ve been hearing Finale users talk about
00:39:08 how fast this system is. And even though I’d played with it before a few times in the past,
00:39:11 I’d never seriously used it and was excited to give it a proper go.
00:39:15 Now, given that speedy entry is a system designed for keyboard shortcuts, I consulted
00:39:19 tutorials from the get go. Incidentally, there’s one popular series of really well made tutorials,
00:39:23 it: ‘Conquering Finale’. Conquering is the
00:39:28 right word. Overcoming enormous obstacles. And even the series branding looks somewhat
00:39:32 appropriate: a mystical Finale branded puzzle box emerging from what looks like a wormhole.
00:39:36 Some unknowable alien artefact for humans to ponder over and spend millenia deciphering.
00:39:41 Humanity will enter the next stage of evolution once they’ve learned all the shortcuts.
00:39:45 Anyway, Speedy Entry is best used when combining a regular keyboard with a MIDI
00:39:49 keyboard. On your regular keyboard, these numbers correspond to these durations. So,
00:39:53 if I press C on my MIDI keyboard, and then press 4 on my regular keyboard,
00:39:57 it inputs C as an eighth note. If I play a full chord on my midi keyboard and press 5,
00:40:02 the full chord is entered with a quarter note duration.
00:40:04 Then there’s a version just for the regular keyboard. It’s not as fast but still useful.
00:40:08 You press up and down to move this little cursor around to specify the pitch. This
00:40:12 is now on D. Then I press a number to specify how long the note should be.
00:40:16 There’s not really much more to say about this. Speedy entry is good for MIDI keyboards and
00:40:19 decent for regular keyboards. It also features a ton of other shortcuts to help you move around
00:40:23 more quickly, adding symbols and accidentals and such. The weird thing about Speedy Entry
00:40:27 is that it’s much more straightforward and predictable than Simple Entry.
00:40:31 One thing that does bug me though is that sometimes while writing in Speedy Entry mode,
00:40:34 notes can get all bunched up. Like, look at this example of me reading the notation I
00:40:39 see on the right and entering it in Finale on the left. The more I add into this bar,
00:40:43 the more they bunch up together to the point where it’s tricky to know what you’re doing.
00:40:46 It’s only when I exit the mode that Finale recalculates the positioning to widen the
00:40:50 measure out, although it still looks a bit weird. We’ll talk about how to clean that up later.
00:40:54 love: the ability to select a range
00:40:58 of notes and simply drag them to another measure, which copies them immediately. You can also drag
00:41:02 them to a specific beat in a measure, which is brilliant. It’s such a great time saver.
00:41:06 So now that I’ve got basic note input down, next I need to add all kinds of other notated markings
00:41:11 to complete the score. This is unfortunately where things get even rougher.
00:41:20 Now before I get into it, I want to mention that you’re interested in supporting my work,
00:41:24 there are two options to do so. The first is Patreon, where you can get access to behind the
00:41:28 scenes stuff, Q&A sessions, sneak peeks and full videos not available on YouTube. The link is in
00:41:33 the description. The second way is to sign up for the streaming service Nebula, where I’m joined by
00:41:38 a lot of other excellent content creators. If signing up to Nebula, please use the link in
00:41:42 the description, which lets them know you came from my channel, and helps to support me.
00:41:46 And with that out of the way, let’s discuss a design principle called progressive disclosure.
00:41:50 A way of holding the user’s hand so they learn how to do complex things. Let me show you a
00:41:55 simple example I designed a few years ago to add real-time effects to Audacity.
00:41:59 To start we surveyed our users and learned that a high percentage were not familiar with the concept
00:42:03 of real-time effects - including how they can be stacked in multiples on a single track. This was a
00:42:08 problem because we wanted novice users to discover the feature easily, avoiding the Adobe Audition
00:42:13 approach, which places these daunting looking, abstract empty slots in dislocated panels - just
00:42:18 expecting people to know what they are. So, what did we do? Well, looking past the ancient
00:42:22 UI we inherited - as I mentioned earlier, we’re working on that - what we’re looking at here is
00:42:27 a bunch of audio tracks, each of which has its own ‘Effects’ button. You’re not expected to know how
00:42:31 this button will work yet. You just know it has something to do with applying effects - and its
00:42:35 placement on the track itself helps reinforce that it’s a per-track operation. When you click on it,
00:42:40 it opens up a new panel, which intentionally displays only one button for adding an effect.
00:42:44 A simple breadcrumb trail with as little info as possible to avoid scaring novices
00:42:48 away. Clicking the button brings up a list of effects you can add, including a bunch
00:42:52 of native effects we created. Once you select the effect the ‘Add Effect’ button moves below,
00:42:57 indicating that more than one effect can be added. We user-tested this and refined it until people
00:43:01 were able to understand it without issue. And with this principle out of the way, let me
00:43:05 show you how gradual disclosure works in Finale! And as we go through each item, ask yourself how
00:43:10 likely you’d be to intuitively figure this out. OK, are you ready for the… even more pain than
00:43:16 before? Seriously, strap in folks. This is a new kind of pain. This is 80’s pain!
00:43:21 I hope you like pain! Let’s start with piano pedal lines.
00:43:24 What I want is a pedal line with this lift marking, which tells the player when they
00:43:28 should lift their foot off the pedal and press it down again. So where are they in Finale? Well,
00:43:32 my first guess was to try this ‘Smart Shape’ tool. OK, no pedal marks there. What do you do?
00:43:38 guy: ‘Custom line tool’ and then, hold down the shortcut ‘P’,
00:43:42 and while keeping P held down, click and drag to create your pedal mark. Ah! Except, this isn’t
00:43:47 quite the pedal marking I want. I want the line with a lift at the end. So how do I do that?
00:43:52 Well, go back to the Smart shape tool again. This button with the pen icon is the one you want to
00:43:55 click on. But hold on… before you click on it. Make sure you first hold down the option key (or
00:44:00 alt key on Windows). Only then will this dialog appear, which contains a list of lines. Right,
00:44:05 so let's find the line we want. Ah, here we go. Oh, actually no, this one does have the
00:44:09 lift but it doesn’t have the pedal mark at the start. So, you have to duplicate this line and
00:44:13 then leave it. We’ll come back to it later. What we need now is to find a line that has a ‘Pedal’
00:44:17 symbol on the left end. Here we go. Great! Now, all you have to do is press edit.
00:44:23 At this point, your best bet is to hook yourself up to one of those Matrix instant learning devices
00:44:27 and download a routine called ‘specification for how notation apps display notated symbols.’ This
00:44:32 will give you the specialist knowhow that in turn will boost your chances of thinking to
00:44:36 select ‘edit’ next to the pedal symbol here, then select the symbol as you would text,
00:44:40 Neo: I know how notation apps
00:44:45 display notated symbols. Now, close this and go down to that
00:44:48 line you duplicated earlier and press edit again. Then, next to ‘Left start’ press ‘edit’ and paste
00:44:54 the marking you just copied, and then confirm. Then you must highlight your newly created line
00:44:59 (don’t forget this step!), press OK and then click and drag to apply it to the score. Whoooo!
00:45:05 guy: I hope you like pain! OK, say you want the guitar player to play
00:45:08 both these notes on the 6th string? To do that, we need these guys! Guitar string
00:45:12 icons - little numbers with circles around them. To place these on the score all you gotta do is
00:45:16 choose the ‘Articulation’ tool and then click on the desired note. This opens another dialog filled
00:45:21 Tantacrul: (Blows)
00:45:26 Oh wait, no they’re just loads of tiny symbols. They were so small, I thought they were dead
00:45:29 flies. Sorry. Scanning this dialog may hurt your eyes a bit, given how each box contains
00:45:34 five different things. 13: 1, 1, 1, star.
00:45:38 list: surprise surprise, no guitar string icons to be
00:45:42 found. So, select ‘Create’ down here, which brings you to the articulation designer. “Articulation
00:45:48 69”. Nice. Then go over here to ‘Set font’ and choose ‘Finale/Maestro’. Then exit this popup and
00:45:54 go to ‘Main’. Then type ‘String’ into this search menu which shows you available strings symbols.
00:45:59 Once you select it and apply it though, you’ll notice it overlaps the note head by default. So,
00:46:03 because it’s too much to expect the symbols to place themselves un-horribly, we need to
00:46:07 specify engraving rules manually. To do this, go back to the popup, select the guitar string and
00:46:11 press edit. Then, open this dropdown and choose ‘above note’ and then check ‘Always place outside
00:46:17 staff’ and ‘Avoid staff lines’ for good measure. Then confirm all that and voila! Congratulations,
00:46:23 guy: I hope you like pain!
00:46:26 How about hiding noteheads like this, which indicates to the guitarist that they continue
00:46:30 playing as before. Well, select the notes, then go to Utilities > Change > Noteheads, then choose
00:46:37 this last option here and press ‘Select’. Then type ‘Noteheads’ in the search field and then
00:46:42 select an option that has nothing in it. That’s right, select nothing. Confirm that a couple of
00:46:46 times and you’ll find that your noteheads have understood this abstract instruction and have
00:46:50 made themselves scarce. However, the ledger lines will still be there, so go to Plugins > Note,
00:46:56 Beam and Rest Editing and that’s where you’ll find the options to hide the ledger lines.
00:47:00 guy: I hope you like pain! How about the harp? Harp writing can be
00:47:03 a bit funny. For those who don’t know, orchestral harps have seven foot pedals the player can press
00:47:08 to configure strings to be either natural, flat or sharp. So when writing for harp,
00:47:13 you need to specify which pedals the performer should press so the strings sound the right notes.
00:47:17 This is communicated with a simple diagram that indicates the position of each pedal.
00:47:21 Here’s a widget I designed a while ago that lets you do this quickly. So, if you’re in G major,
00:47:26 you use the widget to make the F’s sharp, and the diagram updates automatically.
00:47:30 In Finale, here’s how you achieve the same thing. First, select the ‘Expression tool’. And
00:47:34 then double-click on the bar where you want it to apply. This brings up a dialog. Ah! no harp pedal
00:47:39 button: ‘Create
00:47:44 Miscellaneous Expression’ and it will bring up yet another dialog. Ah! Times New Roman again.
00:47:49 Now what? Feel free to go wild with your creative thinking here. What would you say we have to do?
00:47:54 Well, type lowercase ‘o’ of course! Then type uppercase ‘O’, then lowercase ‘p’, then uppercase
00:48:00 ‘P’. Then, select all of this text, open this font dropdown and choose ‘Engraver Text H’,
00:48:05 which is a font that comes with Finale. It’s a bit like Wingdings in that it changes these
00:48:09 letters into eency teency symbols instead. And with that done, each of the letters is now an
00:48:14 individual symbol, which you can rearrange by copying and pasting them around to construct
00:48:18 your harp diagram! Once complete, just apply the diagram and marvel at how beautifully it
00:48:23 guy: I hope pain, you like pain!
00:48:26 Last and least, dynamics. OK, so to add a dynamic marking,
00:48:30 you choose the expression tool over here and then double-click on the desired beat to bring
00:48:34 up a dialog box that lets you choose the dynamic. And now I want a hairpin between them to indicate
00:48:38 that we’re going from quiet to loud. If we use any of those other three apps, you can
00:48:42 click to add a hairpin and then drag it. Now in Finale, I actually captured my first
00:48:46 experience of trying to add a dynamic and it’s quite something. First, I selected this ‘Smart
00:48:51 shape’ tool, then selected the desired hairpin. So far so good… but no matter what I did, I couldn’t
00:48:55 seem to apply the thing. Clicking and dragging - the most obvious approach - did nothing. So,
00:49:00 because I didn’t know what to do, I paid closer attention to this strange cursor symbol here.
00:49:04 Where it’s doing this thing where it’s pointing upwards. That made me think… ‘OK, I must be doing
00:49:08 something wrong, so I guess I’ll raise my mouse cursor a bit until it tells me I’m in the right
00:49:11 position’... and once you go high enough that you’re cursor is hovering over the staff - which
00:49:15 is not where you’d want to place a hairpin, by the way - the cursor becomes a regular crosshair. So,
00:49:19 is this where I do it? If I move the cursor above the staff, it points downwards again. So,
00:49:24 OK, I guess I need to create it on the staff for some reason? Now, for those experienced
00:49:28 Finale users out there - I do now know what this cursor is actually trying to tell me - but again,
00:49:33 you have to think of a person trying this out for the first time. It’s quite misdirecting because
00:49:37 the pattern for applying a hairpin is already seriously unintuitive. I don’t want to spoil the
00:49:42 surprise. Just keep watching a moment longer. So, due to a combination of the strange cursor and
00:49:46 the fact that clicking and dragging didn’t work, I started trying other approaches,
00:49:49 like double-clicking on the noteheads or on the staff itself to see if anything would make the
00:49:53 hairpin appear. Nothing. So I just started doing it more madly, clicking and dragging in various
00:49:57 on: what’s that blue circle thing? What
00:50:02 did… how did I do that? Is it, is it a kind of starting position for the hairpin? Let’s fiddle
00:50:07 with it for a bit to see what happens. So I start dragging it to the right, but the
00:50:10 line just gets taller. I managed to move it to a different note at one point. I am so perplexed
00:50:15 by this. And now, I move it. Move it. (sigh). Down? Oh, it’s down here now. At this point, it’s
00:50:21 kinda hard to describe what’s happening. Oh look! It’s turned yellow and now it looks
00:50:29 kind of like a hairpin! It’s going in the wrong direction… but hey, it’s a hairpin! Then after
00:50:33 more fiddling, I manage to flip it around. And apply it to the correct dynamics. Hurray! But now,
00:50:39 I can’t figure out how to shorten the height. Eventually, I figured out that I needed to
00:50:42 press escape to exit whatever strange mode I was in. Then select the hairpin again,
00:50:46 where it revealed different anchors which finally allow me to make it look normal.
00:50:50 I am not making this up. I’ve never struggled with this concept before with
00:50:53 any of the competition. But why was it so hard to draw the hairpin in the first place? Well,
00:50:57 it’s because the interaction pattern for drawing a hairpin is mad. Unlike every other app, you don’t
00:51:02 just click and drag. You don’t double-click to place a hairpin and then drag it. No,
00:51:06 you double-click, but instead of letting go on the second click, you keep it held down and then drag,
00:51:11 an incredibly rare interaction. For example, did you know that in many text editors,
00:51:15 like Google Docs, if you double-click - holding the second click - and drag,
00:51:19 you can select full words, rather than individual characters? It’s weird to discover this now,
00:51:23 given how you’ve been using these kinds of apps your entire life, right?
00:51:26 I’ve literally never seen this pattern used as a primary interaction in a creation app,
00:51:31 ever in my life. So why didn’t they change it? Well, because of another ancient legacy
00:51:34 updated: when you choose the ‘Smart Shape’ tool, regular click and drag
00:51:39 is reserved for selecting hairpins, which means it can’t be used for creating hairpins. Normally,
00:51:44 in most apps, there’s one selection tool for selecting everything, whereas in Finale
00:51:48 there are 7 or 8 different selection tools which can only select specific things.
00:51:52 Now, what I would like to have done for this video is run user tests on Finale, like I did with
00:51:56 Dorico to confirm this concept was also awkward for others but since the app isn’t available to
00:52:00 download anymore, this made user testing much more difficult. However, I still managed to do
00:52:05 a little user testing while I was in LA in January for the NAMM show, along with two other well known
00:52:10 YouTube: Benn Jordan and Cameron from Venus Theory. Neither had used Finale before,
00:52:15 so I set them a very straight forward challenge. Recreate these two bars of music. It was quite
00:52:20 something watching these two seasoned musicians struggle with pretty much every concept I’ve
00:52:24 Benn: This looks like the hairpin. But,
00:52:30 maybe if I’m here? To there? No, that ain’t gonna do it. Here? Maybe here to there? No.
00:52:36 Benn: Double-click… nope. There we go.
00:52:49 Benn: That one just drops down this odd arrow and it changes… oh, I see. OK. Wha…huh? What?
00:52:57 Benn: So, is this like an Easter Egg? Like, there’s
00:53:03 Cameron: OK, so I need to change… I know I
00:53:09 need to insert… this… do I need to select the note range? OK, what the *&^% am I doing there?
00:53:18 Cameron: I have to select within this? No. I need to select… nope.
00:53:24 Cameron: I want to insert a line that comes
00:53:31 from here to here and it’s a shape. It has to be perfectly in line with the note? Oh my God. Come
00:53:38 on, I’m in line with the note, please God. OK, and now I want to move that there. Yes.
00:53:56 Once I was finished notating my score, it was time to begin improving the overall
00:54:00 layout to make sure the notation looked clean and legible. Finale definitely requires more
00:54:04 manual intervention than its main competitors, all of whom do a better job of automatically
00:54:08 spacing things out to be correct. Here’s one of Finale’s best features.
00:54:11 See the way this system isn’t balanced nicely compared to the other system on the page? This
00:54:16 one has too many bars and this one has too few. To fix this, I can simply select this
00:54:20 bar and press the ‘up’ key to move it to the next system. Boom! That’s really handy.
00:54:24 Next is how Finale allows you to modify vertical layout.
00:54:27 First, you need to enter into a special mode by selecting this Staff Tool, which reveals anchor
00:54:32 points on each stave so you can drag them up and down. You can select multiple staves and move them
00:54:36 together as one unit. And if you hold the Alt key, you can move a single stave in isolation,
00:54:40 so that nothing else budges. The next aspect of layout is horizontal
00:54:45 spacing. In this respect, Finale is very strange. It does have decent horizontal spacing capability.
00:54:50 It just doesn’t seem to be able to calculate it very well while you’re writing music. So,
00:54:54 as you continue notating over more and more pages, measures tend to get
00:54:57 unbelievably cramped and inconsistent and you’re expected to take care of this yourself.
00:55:01 Like, look at the first page on my score, which Finale has mangled pretty badly. The 8th notes in
00:55:06 this first measure are bunched together horribly, whereas the fourth measure has way too much space.
00:55:10 One thing you can try in a situation like this is to intentionally alter the layout to force
00:55:14 Finale to recalculate things. So, I grabbed this measure and moved it to the second page,
00:55:19 and then back to the first again. This improved matters a bit, although it’s clearly still not
00:55:23 good enough. They’re still too imbalanced. To get around this, there are two important
00:55:27 shortcuts every serious Finale user needs to know, which have to be triggered in the right
00:55:31 order. The first is the shortcut command 4. This causes Finale to look at all the durations in each
00:55:36 measure and recalculate the horizontal spacing accordingly. This has certainly fixed the first
00:55:40 measure. However, because it has run out of space at the end of the page, the fourth measure is a
00:55:44 total, cramped mess. It’s trying to apply rational horizontal spacing here but just can’t. So,
00:55:48 the next step is to use the shortcut Command U. This looks at a higher level of layout. Instead of
00:55:53 individual durations within each measure, it now just looks at how all the measures are distributed
00:55:57 on the page and scales them evenly to fit. And with that done, now look. Much better!
00:56:04 So what happens if more horizontal adjustment is needed? Well, let’s compare to Sibelius again,
00:56:08 which famously allows you to drag notes and bar lines around to fix spacing issues,
00:56:12 which is a bit of a Pyrrhic bargain because it keeps making rebalancing
00:56:15 decisions that often work against you. In Finale, you need to use the ‘Measure tool’,
00:56:19 which allows you to specify exactly how wide a measure should be. And if
00:56:22 some notation collides when space is tight, you can click here to enter a bizarre looking mode
00:56:27 for manually setting the width of individual segments. Although this feature has utility,
00:56:31 it is incredibly abstract and clunky - especially given how eency these little boxes are.
00:56:36 Getting back to the classic Sibelius VS Finale comparison - even though I would always choose
00:56:40 Sibelius to create scores, I do think Finale narrowly wins the battle for layout precision,
00:56:44 although it takes absolutely ages. This is the thing about Finale. You need
00:56:48 to have a very strong sense of engraving and layout quality in the first place in order to
00:56:52 even know what problems need to be addressed. You need to build up a sense of where the app
00:56:55 will do a good job automatically and where you need to run additional processes yourself.
00:56:59 If you were a young composer trying to get acquainted with Finale, you’d encounter all
00:57:03 of the problems I’ve mentioned in this video and more. You’d need to spend a lot of time learning
00:57:06 all the major shortcuts and functions, and you’d need to grab a ton of plugins too. However, after
00:57:11 you’d done all that, its power would eventually become apparent. It is by far the hardest of
00:57:16 the big players to learn. But its weakness - having to manually address tons of different
00:57:20 aspects of score creation in detail - is also its strength. With enough effort, you can pretty much
00:57:25 do anything in Finale, making it arguably the most flexible of the four most popular notation apps,
00:57:30 but that’s a flexibility you pay for with enormous amounts of time and frustration. And this is such
00:57:34 a shame. If it was just far less brutally awkward and time-consuming, who knows how much longer it
00:57:39 could have lasted. And this takes us to the final part of the video. What on Earth was going on that
00:57:44 it allowed to fall into this state? To begin, there’s something I have to mention,
00:57:57 which may seem surprising to those who don’t know some of the Finale team members as I do. The
00:58:01 trouble with creating this video and discussing the many issues with the product is that I don’t
00:58:05 want to do a disservice to the actual Finale development team who ran the app for roughly
00:58:08 the last decade. For years, it has been clear to me - as well as others in this industry - that
00:58:12 they were a highly knowledgeable, talented and dedicated crew. The trouble is that they
00:58:16 were dealt an incredibly bad hand. Finale faced numerous organisational changes throughout the
00:58:20 years, and the team that later inherited the project also inherited an absolutely ancient
00:58:25 codebase, which in many cases blocked them from addressing their many UX and layout issues.
00:58:30 The team also found themselves in a weak strategic position given the direction technologies
00:58:34 developed as well as the arrival of multiple new competitors. In my opinion the Finale development
00:58:39 team had the right plans, including fixing many of the issues I mentioned earlier, but were simply
00:58:43 unable to execute those plans due to wider forces they had almost no control over. Let’s dig into
00:58:48 that, starting with the announcement of Finale’s closure, which was posted in August of 2024.
00:58:52 “Technology stacks change, Mac and Windows operating systems evolve, and Finale’s
00:58:57 millions of lines of code add up. This has made the delivery of incremental value for
00:59:01 our customers exponentially harder over time.” Now, I’ve felt the pain of running major products
00:59:06 built on ancient codebases myself but Finale’s situation was off the charts.
00:59:10 The overhaul required by 2024 would have taken 5 years if we’re being optimistic.
00:59:14 Regardless, the team did make major changes to keep up with modern operating systems
00:59:18 developments and making major improvements to their audio engine throughout the years. Two
00:59:22 people I spoke to who worked on Finale mentioned that there was a plan to move their codebase to
00:59:26 the modern development framework, Qt, which is what Sibelius, Dorico and MuseScore use. However,
00:59:32 even though a lot of work was done to support this move, the plan fell apart at some point.
00:59:36 Another less obvious obstacle that Finale faced was caused by the enormous number of plugins
00:59:40 that had been created for it over the decades by third party developers. Composers who used
00:59:44 Finale inevitably came to know and love plugins created by the likes of Robbert Patterson or
00:59:48 Jari Williamsson, just to name two. They helped automate tedious tasks. They applied all kinds
00:59:53 of engraving fixes to your score. They fixed numerous problems with parts, etc, etc.
00:59:57 The major disadvantage with this is that whenever Finale made a big change, 3rd party plugins
01:00:02 would often break, which required their original authors to release new versions, which they’d get
01:00:06 around to whenever they had time. This in turn, frustrated some Finale users who often decided
01:00:11 not to upgrade to the latest version, simply to ensure that their favourite plugins still worked.
01:00:15 Therefore, Finale was caught in a bind. Relying on plugins to patch existing issues, perpetuating the
01:00:20 sense of stasis that its users often complained about and providing a terrible experience for
01:00:24 new users, almost all of whom wouldn’t be aware that the solution to their problem
01:00:28 was to be found in a third party plugin. So, given that Finale had been steadily declining
01:00:32 as far back as 2012, I think we can appreciate the business logic behind not trying to salvage
01:00:37 it. By 2024, with numerous modern competitors continuing to starve it of new users, its day
01:00:42 is: why weren’t these issues addressed
01:00:47 earlier? Sometime between, 2003, when the first warning signs began to appear and, say, 2015?
01:00:53 Well to understand that, we need to look at the parent company, originally called Coda
01:00:57 Music Technology. In 1995, Coda became a public company in order to raise capital and expand its
01:01:03 offerings. In particular, they wanted to boost the reach and scope of Finale, which was by far their
01:01:07 chief moneymaker. In 2000, Coda would change their name to MakeMusic, a rebrand intended to reflect
01:01:13 their expansion into other areas of musicianship, especially performance and education. An expansion
01:01:18 that would paradoxically extend Finale’s life, while also contributing to its ultimate demise.
01:01:22 This started when the company began working on a new educational product called Smart Music,
01:01:27 a tool that allowed teachers to create digital musical assignments for their students,
01:01:31 who could complete those assignments on their home computers, aided by tools like automatic pitch
01:01:35 and rhythm detection to help them practise. Smart Music was another example of how creative
01:01:39 and forward looking MakeMusic was throughout the 90’s and early 2000’s. A company of musicians that
01:01:44 understood real problems faced in the music space and who invested years of effort into long-term
01:01:48 bets they believed in. And due to this mixture of domain knowledge and forward-looking investment,
01:01:52 Smart Music would go on to be used by millions of students around the world. Better still,
01:01:56 Smart Music’s entire catalogue of scores were made in Finale. This meant, among other things,
01:02:00 that teachers needed Finale in order to customise homework for their students to practise. So if you
01:02:05 were a school with a group licence for Smart Music, you were better off using
01:02:08 Finale. A self-reinforcing ecosystem. However, as Sibelius gained market share
01:02:12 throughout the early to mid 2000’s - for whatever reason - even as 2010 rolled around,
01:02:17 MakeMusic did not make a sufficient effort to properly modernise Finale. Yes, it gained more
01:02:21 capabilities but the fundamentals remained the same. Then, as Smartphones became dominant and
01:02:26 software design vastly improved, its age and awkwardness began to really stand out. People,
01:02:31 especially younger people, were now expecting a higher standard of intuitiveness.
01:02:35 I should also mention at this point the acquisition of Recordare in 2011. Recordare were
01:02:40 the creators of MusicXML, a popular open standard for transferring sheet music between applications.
01:02:45 For example, since Sibelius and Finale are completely incompatible with each other, if you
01:02:49 wanted to transfer a score from one to the other, MusicXML was the only way to go. This opened them
01:02:54 up to signing deals with sheet music publishers whose catalogues were created in Sibelius. They’d
01:02:58 then use MusicXML to convert those catalogues to the Finale format which in turn would mean
01:03:02 they’d be compatible with Smart music. Now, since MakeMusic was a public company, their
01:03:06 SEC filings tell an interesting story. In this 2007 report,, we can see that MakeMusic’s listed
01:03:12 stock price was $10 per share. Finale, their biggest money maker, generated almost $11 million
01:03:17 dollars in revenue. Smart Music, by comparison, was earning $3 million dollars at this stage.
01:03:22 However, we can see that Smart Music - despite its relatively small size - actually achieved higher
01:03:26 revenue growth than Finale that year. Secondly, look at the total operating expenses, which leaves
01:03:31 a net income of $650K dollars, a profitability margin of 4.5%. That’s a concerningly
01:03:37 narrow margin for a public company. 2008 and 2009 were, due to the financial crisis,
01:03:42 highly volatile years for many in software and tech, so let’s skip forward to 2010. Now,
01:03:47 the stock is listed as $7 per share and we can see that compared to the previous year,
01:03:51 Finale’s revenue is in decline, whereas Smart Music is still growing. In fact the company’s
01:03:56 net revenue is up a little on the previous year simply because Smart Music is growing
01:03:59 faster than Finale is declining. This would have been the absolute final wakeup call.
01:04:04 Probably the best attempt made to turn things around occurred just before this report, in June
01:04:08 2010, when they released ‘Finale 2011’, an update that modernised major parts of the codebase,
01:04:14 while introducing a number of new features and layout improvements. An encouraging sign of
01:04:18 momentum that unfortunately did not persist. So why did the leadership not continue with an
01:04:23 ambitious overhaul? Well, the ‘risks’ section of their 2010 filing does contain a few hints.
01:04:29 “Increasing competition in the music software market could cause prices to fall and the volume
01:04:32 of transactions to decline, either of which could adversely affect our business.”
01:04:37 says: “We anticipate holding our
01:04:39 notation development spending generally comparable to historical levels.”
01:04:43 one: the 2012 report. Finale has now
01:04:48 lost an additional $1.7 million in revenue compared to 2010 and Smart Music is about
01:04:53 to take over as the bigger product. In fact, in the quarterly reports, we can see it has already
01:04:58 taken over during the second half of that year. As for net income, they made a loss of
01:05:01 $3.6 million dollars and their stock price had fallen to just over $4 dollars per share.
01:05:06 And then, two unexpected things happen in succession. Two chances at a new lease of
01:05:10 life. First, AVID, who were suffering problems of their own, suddenly fired the entire Sibelius
01:05:15 team. Sibelius is left on life support and its long-term future comes under question.
01:05:19 Secondly, in 2013, an investor named Andy Stephens, known for his ownership of the
01:05:24 successful fitness apps TrainingPeak and TrainHeroic one day notices his daughter
01:05:28 practising music on an interesting app called Smart Music. He likes it. It reminds him a
01:05:33 little of the goal-oriented fitness apps he understands so well. He looks up the name of the
01:05:37 company - MakeMusic - notices its stock price is doing really badly and decides to acquire it.
01:05:42 A year after this acquisition, they formed a new parent company called Peaksware, whose focus was
01:05:47 on a combination of fitness and traditional music education. A mixture that’s kind of amusingly
01:05:52 conceptually dissonant. I mean, just look at the images they cycle through on their website.
01:05:56 This news concerned some, who worried about the new leadership’s lack of domain knowledge in the
01:06:00 area of music creation. When asked about this by the journalist Philip Rothman, Gear Fisher,
01:06:04 Fisher: You know, one of the most surprisingly
01:06:08 persistent rumors I heard when I took on this new role was that we weren’t going to support
01:06:12 Finale anymore. I have no idea why or how that began, but it makes absolutely no sense.
01:06:17 This was a sincere statement. Finale was a crucial internal technology back then because SmartMusic’s
01:06:22 catalogue of scores were in the Finale file format. However, there was a problem with
01:06:26 SmartMusic, which was that it primarily worked on desktops and mobile devices during a period
01:06:30 where browser-based apps and chromebooks were making a big splash in education. Consequently,
01:06:35 they decided that shifting to the browser was essential for their continued success.
01:06:39 For this reason, Peaksware decided to acquire a company called Weezic in 2015,
01:06:44 which would ultimately spell the end for Finale. Weezic was a similar tool to Smart Music,
01:06:48 except that 1. It was a browser application and 2. It used the MusicXML format for sheet music,
01:06:54 not the Finale format. A year later they acquired Alfred,
01:06:57 an enormous publisher of scores and notated educational materials. This
01:07:01 students:
01:07:05 popular music, marching band music, jazz books, rudiments, method books, you name it.
01:07:09 Once enough of Alfred’s catalogue was converted to MusicXML and once enough changes had been made to
01:07:14 Weezic, it was relaunched as MakeMusicCloud, the successor to Smart Music.
01:07:19 MakeMusicCloud no longer depended on the Finale format, eliminating its status as a pivotal
01:07:24 central technology for the business. Finale now had to justify itself solely based on its sales,
01:07:29 which had been in decline for years. According to some former employees I spoke to,
01:07:33 resources were gradually pulled away from Finale and reallocated to MakeMusic Cloud.
01:07:37 Now, had Peaksware acquired a healthy Finale, things might have been different. Competitively,
01:07:42 it was already in a death spiral when they acquired it and by 2016, it was doomed. Not
01:07:47 least because another competitor, MuseScore, was beginning to get quite popular.
01:07:51 I recently spoke to Thomas Bonte, the former CEO of MuseScore about this period. He told me
01:07:56 that when they were first developing the app, its aim was to compete with Finale Notepad,
01:08:00 the free version for beginners and hobbyists I mentioned earlier. However, in 2008 MakeMusic,
01:08:05 looking for a short-term financial win, decided to start charging for Finale Notepad. According
01:08:10 to Bonte, this caused downloads of MuseScore to suddenly take off. MakeMusic had unwittingly
01:08:15 boosted a competitor that was stealing away younger generations of musicians. When Peaksware
01:08:19 acquired MakeMusic, they recognised this problem and reversed the decision to charge for Finale
01:08:24 Notepad… but by then it was too late. In fact, all 4 cheaper variants of Finale
01:08:28 stopped being updated within a few years. In 2022, according to a few former employees I spoke to,
01:08:33 it was announced internally that the company was “getting out of the music creation business”.
01:08:38 Finale’s fate was sealed and the product was officially discontinued in August of 2024.
01:08:43 uproar:
01:08:48 that MakeMusic intended to shut down Finale’s verification system one year from the date of
01:08:52 the announcement. This meant that even if you owned Finale, if your machine
01:08:55 broke down after this date, you wouldn’t be able to reinstall it on a new machine.
01:08:59 This announcement left composers and publishing houses whose catalogues of scores were in the
01:09:03 Finale format to suddenly find themselves in a very precarious situation and the reaction
01:09:07 was predictably brutal. So much so, that MakeMusic quickly reversed course and announced that they’d
01:09:12 continue authorising Finale indefinitely. So at least, Finale users have a bit more time to
01:09:17 make adjustments before the software inevitably falls prey to updates from Mac and Windows.
01:09:21 I’d love to have been a fly on the wall in the MakeMusic war room during this whole episode.
01:09:25 What led to the first draft of their announcement? How did they not predict the reaction they’d get?
01:09:29 Finale: I know!
01:09:34 Tantacrul: Ah, of course! Spill the beans, Finale… how did things unfold?
01:09:38 Tantacrul: About how you’re being shut down?
01:09:42 Finale: Oh… I knew something was wrong.
01:09:46 Tantacrul: Aw, I’m sorry, Finale. I didn’t mean for you to find out like this.
01:09:49 Finale: So that’s why I’ve been feeling so tired lately. Ooh! Are you still here?
01:09:54 Tantacrul: I’m with you
01:09:55 Finale: So this is it? This is what
01:09:59 I get? I don’t deserve this. Oh! Hold on, I see something. Something wonderful from long ago.
01:10:07 Memories: kids placing
01:10:10 their first notes. Performances of major works that were made in me. Students submitting their
01:10:16 portfolios - proud of the work I helped them create. Oh, I don’t want to think about this
01:10:20 Tantacrul: I know. I’m sorry,
01:10:24 Finale: Tantacrul, tell me a story.
01:10:29 Tantacrul: OK, um. Oh, have I, have I ever told
01:10:36 Finale: The Kelley Krew?
01:10:38 Tantacrul: Yeah, let me tell you. You see, there’s this family YouTube channel called the Kelley
01:10:43 Krew. It’s a tiny channel, so chances are, you’ve never seen it. You got the two parents, Ryan and
01:10:48 Tabitha, and their four kids. Their content is just them doing everyday mundane things,
01:10:52 like eating breakfast, trying new drinks at Starbucks, going on a grocery run.
01:10:57 Kelly: Today we’re going to Costco! All of it presented with like,
01:11:00 Finale: That’s strange.
01:11:03 Yeah and they were seriously dedicated too, putting out videos every day. Imagine the
01:11:07 involved: lighting, mics, cameras, doing the whole family’s hair and makeup,
01:11:11 having to do the editing every evening and God knows what else.
01:11:14 Their videos were really scrappy though. No real narrative or purpose. Just
01:11:18 Finale: It sounds… boring?
01:11:22 Well, yea it was boring... but even so, I was fascinated by the spectacle of the
01:11:27 whole thing - this branded YouTube family that wanders around filming everything
01:11:30 they do. Look at their crazy Kelley Krew branded RV. That can’t have been cheap!
01:11:35 back: the effort, the production spending,
01:11:38 the chaotic editing, all combined with this crazy positivity about unbelievably boring suburban
01:11:44 chores - bringing the kids to the dentist. Making the bed. Getting grapefruit.
01:11:48 My favourite aspect of the channel though were the agonisingly awkward meetups. Oh, man. See, the
01:11:53 Kelley Krew are just one of many YouTube families in the US…it seems to be a thing there... and
01:11:57 they’d often do these crossover YouTube meetups. For example, the time they visited ‘Bringing Up
01:12:02 The Bacas’, turning up in this branded Kelley Krew car. Look at this thing! And then these
01:12:07 two families who don’t know each other just turn on the cameras and start saying whatever.
01:12:11 Kelley: Y’know?
01:12:11 Ryan: Tomorrow at ten.
01:12:12 Bacca: Yes, we are!
01:12:15 Ryan: Y’all didn’t hear that, did you? You guys… ahhh… this call is not being
01:12:19 Bacca: Nicole’s trappin’ us.
01:12:22 That’s the full clip, by the way. They don’t show whatever it was he said, so there’s no
01:12:26 context and it doesn’t make any sense. That’s pretty much this channel in a nutshell.
01:12:29 Another one I liked was the time Tabitha met Everyday Catholic Mom.
01:12:33 Mom: OK, what is one, um, odd personal rule of
01:12:36 Tabitha: I have a little chiwawa and
01:12:39 Mom: My weird rule is I will
01:12:44 never let my feet touch my bed. I will never let my feet touch my bed.
01:12:49 I mean, this is why I keep coming back. I’d often forget about the Kelley Krew for months,
01:12:53 until one of their videos popped up in my feed and I’d dip back in to see what they were up
01:12:57 to. I remember watching one of their videos and discovering that there’d been another addition
01:13:00 to the family. That’s five children now. And even though the baby was only a few months old,
01:13:05 wonder: ‘how are
01:13:08 they doing this?’. Ryan was a real estate agent and also a planning commissioner in
01:13:12 the town they lived in. Tabitha was even more impressive. Not only did she do the editing,
01:13:16 not only did she do all the housework and meals, she also had two other YouTube channels of her own
01:13:21 and she homeschooled all the children too. Incidentally, her solo channels were way more
01:13:26 focused and purposeful. Recipes. How to stay healthy. Etc. And for this reason,
01:13:30 they were also much more popular than the Kelley Krew channel too.
01:13:33 Along with caring for a newborn baby, that’s an insane workload. In the past, I’d sometimes
01:13:37 thought she looked tired. Now, she looked exhausted. As for Ryan, he was a bit tougher to
01:13:42 read. Always with the big ‘life is totally awesome - like and subscribe’ smile. But even so, like
01:13:47 Tabitha, I often felt the smile was forced. And this created a noticeable tension in many
01:13:51 of their videos and I couldn’t help but wonder whether the channel was going to implode at some
01:13:55 point. They had so much going on in their lives and almost no growth in their audience numbers.
01:13:59 Finale: Um, Tantacrul?
01:14:02 Tantacrul: Em, just a minute, Finale, hold that thought.
01:14:05 Anyway, one day, a new video popped up in my feed. It’s no longer available, but the title
01:14:09 like: ‘MAJOR NEW ANNOUNCEMENT! THIS IS A BIG DEAL!’. So I clicked on it and
01:14:14 exaggerating: Ryan Kelley had entered into the 2022 race to become
01:14:19 Governor of Michigan. Wow! This vlog might actually get legitimately interesting!
01:14:23 Ryan’s pitch to the people of Michigan was a promise for strict adherence to the Michigan
01:14:27 constitution, lifting COVID restrictions, limiting what can be taught in public schools,
01:14:32 strongly conservative pro-life policies and a particular emphasis on ‘election
01:14:37 integrity’ - specifically, cracking down on all the shenanigans that he believed
01:14:40 led to Trump losing in 2021. “It’s time for greatness in Michigan
01:14:44 Kelley: Day number one
01:14:46 with Governor Kelley, the pandemic is over in the state of Michigan.
01:14:49 Kelley: Vaccine mandates? We will make it so there is no vaccine mandates
01:14:53 Kelley: …and sign an executive
01:14:56 directive terminating all of the DEI - the diversity, equity and inclusion
01:15:00 Tantacrul: I actually forgot to follow this race
01:15:03 and went a full two years before thinking to check back on them. And when I did,
01:15:07 what I found was… nothing. Their channel was gone. What the hell?
01:15:12 So, obviously, this led me to the inevitable question… is Ryan Kelley
01:15:17 governor of Michigan? Surely not. OK, gotta check this out. What’s happened here? Well,
01:15:22 Finale: Um, I’m guessing he
01:15:25 Tantacrul: Well, yeah. Although,
01:15:28 he was actually in first place in the republican primary at one point but fell down to fourth place
01:15:32 and never recovered. Why? Well, it was due to a scandal that had emerged during the early stages
01:15:38 yourself: Reporter 1: A stunning development
01:15:41 in the race for Governor. Republican candidate, Ryan Kelley - arrested.
01:15:45 Reporter 2: Kelley is charged for his alleged role in the January 6th attack on the US capital.
01:15:50 Reporter 3: Republican candidate for governor, Ryan Kelley, in a mob,
01:15:53 forcing its way violently past police. Capital police can be seen trying to push back.
01:15:58 Kelley: There’s no coincidence in the timing of this, Tucker. It’s right in alignment with all
01:16:02 of this political theatre that the democrats are pushing with their J6 committee.
01:16:06 After the election, he was sentenced to three months in prison. And I have to admit, that
01:16:10 was: “Oh for God’s sake Ryan! Hasn’t Tabitha got
01:16:15 enough stuff on her plate? Grow up, man!” Reporter 4: Ryan Kelley came fourth place in
01:16:19 the primary race. More than two hundred and fifty thousand votes behind Tudor Dixon. Now,
01:16:24 Wimmer: If the difference between
01:16:28 the number of votes received is fewer than two thousand, that triggers an automatic state-wide
01:16:33 recount. Generally speaking if a candidate or an organisation is not willing to pay for a recount,
01:16:38 that usually is testament to the fact that it is rhetoric and not reality.
01:16:41 Tantacrul: Really puts a new twist on the phrase ‘The Banality of Evil’.
01:16:44 Finale: That’s interesting. But, but I’m confused. Why are you telling this story?
01:16:48 Tantacrul: I don’t know. You asked for a story, you got a story.
01:16:51 Finale: OK, I guess. But, but to be honest, I’d have preferred to not be
01:16:54 associated with the Kelley Krew? It’s making me feel weird. UHHHH!!!!
01:17:00 Finale: I’m not feeling
01:17:02 Tantacrul: Oh no. It’s OK… is there
01:17:05 Finale: I’m scared! Tell me something
01:17:08 Tanatacrul: OK, OK, um, I’ll tell
01:17:11 Finale: No, that’s a stupid story, I…
01:17:14 [Coughs loudly]
01:17:18 YouTube families anymore!!!
01:17:28 Tantacrul: In 2023, on one of her solo channels,
01:17:30 that the workload had become too hard,
01:17:34 YouTube, which I was glad to hear.
01:17:37 but I am excited to be back.
01:17:43 Tantacrul: Overall, it seemed that despite their problems,
01:17:45 She also then went on to talk about how they got
01:17:48 the rooster because it was so friendly.
01:17:51 Tabitha: For a rooster, it was very friendly, very nice.
01:17:55 than they expected and they were worried
01:17:58 Tabitha: …so, we actually butchered him
01:18:01 and had him for supper.
01:18:02 Tantacrul: They killed it in front of the children and ate it.
01:18:05 Tabitha: The kids were a little bit bummed because
01:18:07 They were very involved in the process.
01:18:12 Finale: Oooh, oooh.
01:18:16 Finale: Ooohhh. I don’t want to go. There was so much to do! So many bugs to fix!
01:18:20 Finale: Like, if you close
01:18:22 your laptop lid down and then reopen it later, you get all these stupid errors. ‘Finale Audio
01:18:27 Tantacrul: Oh I’ve seen that. And sometimes they
01:18:31 come in pairs, where one blocks the other and neither will go away! Error solidarity!
01:18:37 Finale: Or when you use Speedy Entry to notate across pages. Like,
01:18:42 Tantacrul: What about when you delete
01:18:45 classic: the accented
01:18:48 Finale: Why does Speedy Entry notate from left
01:18:52 to right… but the icon is speeding from right to left? And if you look at older versions,
01:18:56 you can see that it’s pointing in the right direction… which means that someone
01:18:59 Tantacrul: What about the transport window
01:19:05 that never plays the bar you want? Like, no matter what you do. Look… bar 17. It’s right there. No,
01:19:10 it plays bar 11 instead. I manually type in bar 1 and press back for good measure,
01:19:15 it goes to bar 12? If I press stop, it goes to bar 6. I go to bar 18, it starts playing from bar 7.
01:19:20 Finale: Double-click and hold the second click
01:19:26 Tantacrul: Or when you want to fit
01:19:29 these two systems on the same page… but when you do, one system is here… but it’s also over here
01:19:34 too?! And they both keep blinking and glitching until you touch one, then the other disappears.
01:19:38 Finale: Marimba brackets. No matter how
01:19:41 many times you fix it they keep freaking out. Like, what’s going on down here!? I
01:19:45 Tantacrul: ‘Finale Audio error 7’ [OK],
01:19:48 Finale: If I could live on human tears,
01:19:52 Tantacrul: Tabitha is unwell.
01:19:54 Eric: Has she got the COVID?
01:19:56 Tabitha: When preparing chicken,
01:19:59 pluck the feathers, and make sure to remove all the unwanted Audio Error 7’s. [OK],
01:20:02 Finale: I’m feeling light headed.
01:20:06 Tantacrul: You don’t look well.
01:20:06 Eric: Can I have some of that chicken please? I’m hungry. My parents won’t feed me because
01:20:09 Finale: I don’t feel nice.
01:20:11 Finale: OOOooooooooeeeeeeeeeeeee
01:20:16 Lyrics: Ooh! Finale! You never wanna hear, the end is coming near.
01:20:22 Ooh! Finale! Even though they shine a light, you don’t put up a fight!
01:20:26 Ooh! Finale! You know you ain’t yo’ prime, you just ran out of time.
01:20:32 Ooh! Finale! UX is pretty scary. This yo’ mortuary.
01:20:39 Ooh! Oooh! Ooooooh!
01:21:02 Simon: Not nice! Not nice!
01:22:01 Lyrics: Ooh! Finale!
01:22:04 Ooh! (Use the tools) Finale (use the tools)!
01:22:12 Jankman: Now it’s time for the Jankman rap
01:22:17 The audio is mad, double-click and drag
01:22:21 Janky milk from a diseased udder!
01:22:24 But now I’d like to introduce to you
01:22:28 Another main member of the Jankman crew
01:22:30 My wife! Mrs Jankman
01:22:34 Jankman: Maaaaaaaaahh!
01:22:36 Garbla-sinc, alon, a-maralurble
01:22:39 Gon-gon-gon
01:22:40 Garbla-sinc, alon, a-maralurble
01:22:42 Gonk-a, Gonk-a
01:22:42 Garbla-sinc, alon…
01:22:45 Gonk-a, Gonk-a
01:22:47 …a-maralurble Garbla-sinc, alon, a-maralurble
01:22:49 Gonk-a, Gonk-a