Okay, here is the abstract and summary for the provided transcript about the "Oz" color display principle.
Abstract:
This paper introduces "Oz," a novel principle for color display based on stimulating individual photoreceptor cells (cones) on the human retina at a population scale using precisely targeted laser microdoses. Theoretically, this method bypasses the natural constraints of cone spectral sensitivities, potentially enabling the perception of novel colors outside the standard human gamut, such as by activating only Medium-wavelength (M) cones. The researchers developed a prototype system using Adaptive Optics Scanning Light Ophthalmoscopy (AOSLO) to track eye motion and deliver targeted 488nm or 543nm laser light to thousands of pre-classified L, M, and S cones at 4° eccentricity. Color matching experiments confirmed that attempting to stimulate only M cones produced a perceived color ("olo") lying beyond the natural gamut, described by subjects as an exceptionally saturated blue-green. Further experiments demonstrated that subjects could perceive and interpret images and videos rendered using these Oz colors, distinguishing shapes and motion based on hue information that was absent in control conditions where laser targeting was intentionally jittered. These results serve as a proof-of-principle for programmable, cell-by-cell control of visual perception and open new avenues for vision science research.
Oz: Achieving Novel Colors by Stimulating Individual Retinal Cones
Introduction - The Oz Principle: Introduces a new color display method called Oz, which aims to control vision by directly stimulating individual L, M, and S cone photoreceptors on the retina using targeted light, bypassing the limitations of natural light mixing.
Fig. 1 - System Overview: The prototype uses AOSLO, pre-classified cone maps, real-time infrared eye tracking (at 60Hz), and visible laser microdoses (105k per second across ~1000 cones) modulated to achieve desired activation levels for displaying images/video within a 0.9° field of view at 4° eccentricity.
Fig. 2 - Theoretical Gamut Expansion: Theoretically, Oz could produce colors spanning the entire L/M/S cone activation space. In practice, light leakage ("fractional leak") shrinks the achievable gamut towards the stimulating laser's natural color. The model suggests ~1/3 of light hits the target cone, yet this is sufficient for out-of-gamut colors. Gamut shape also depends on the stimulating laser wavelength.
Design of Prototype - Technology: Built on AOSLO, using infrared (940nm, 840nm) for wavefront sensing/imaging and visible lasers (488nm or 543nm) for stimulation. AO corrects aberrations; FPGA/GPU handle real-time tracking (0.8ms latency) and stimulus delivery, compensating for eye motion and system latencies. Cones are pre-classified using AO-OCT.
Results: Color Matching (Fig. 3): Formal experiments measured Oz colors. Subjects matched Oz color squares (using 488nm or 543nm stimulation) with controllable colors (RGB projector or tunable laser + white).
Key Takeaway: Attempting M-cone-only stimulation (target: olo) produced a color requiring desaturation with white light to match the most saturated natural teals (monochromatic light ~501-512nm), proving "olo" is beyond the natural gamut.
Observation: Olo was qualitatively described as highly saturated blue-green (rated 4/4 saturation).
Control: Jittering microdose targets by ~2 cones caused perceived colors to collapse back to the laser's natural hue, confirming the need for precise targeting.
Results: Image/Video Recognition (Fig. 4): Subjects viewed equiluminant red (L-cone activation) shapes on an olo (M-cone activation) background using a 543nm laser.
Key Takeaway: Subjects successfully identified line orientation (4-AFC) and motion direction (2-AFC) using only the Oz-generated hue cues.
Control: Performance dropped to chance level under the jitter control, indicating hue information was successfully rendered by accurate Oz stimulation.
Discussion - Spatial vs. Spectral Metamerism: Oz represents "spatial metamerism" (controlling spatial light patterns) rather than traditional "spectral metamerism" (mixing spectral primaries like RGB). It can create multiple colors from one laser and colors impossible via spectral mixing.
Discussion - Challenges & Dynamics: Current limitations include small field of view (0.9°), need for fixation, and technical scaling challenges. The system must dynamically update cone stimulation during eye movements to avoid fading (Troxler effect) and render complex imagery.
Discussion - Future Potential: Oz provides a new platform for vision science to probe color perception thresholds, ganglion cell responses, visual illusions, plasticity (e.g., simulating new cone types), and potentially test corrections for color blindness or simulate tetrachromacy.