I think I might have come across another cause of eyestrain, I'll dub as "relative spectral peakiness". It is basically the relationship of a light sources maximum peak amplitude on it's spectral curve relative to its overall average amplitude across its entire spectrum, the narrowness of the peaks, and where on the spectrum the peaks are located. Now I don't have any formal education or much insight into how the eye goes about detecting color and brightness, and how the receptors respond to high amplitude narrow peaks vs lower amplitude broad spectrum humps/smooth curves so all this is just conjecture.
I recently bought a waveform lighting d65 northlux shop light that's 6500k, flicker free, 95cri. It gives me eyestrain, even when positioned to eliminate glare and illuminate at a lux level high enough to be in the pleasing white range according to the kruisthof curve. I also have an original ideal-lume standard t5 fluorescent bias light. It is 6500k, 90cri, and I have confirmed it flickers through slow motion capture on my iphone. Yet it actually causes less eyestrain than the waveform lighting shop light despite being objectively inferior. I think this could be caused by the narrow 450nm peak shown in the spectral measurement of the waveform shop light. Its amplitude is disproportionately high relative to the rest of the spectrum. On the other hand, the spectral curve for the ideal-lume standard is much more concentrated in three distinct peaks of the red, blue, and green phosphors. This less full spectrum inevitably creates a lower cri, but those spectral peaks are much less disproportionate in relative amplitude to each other compared to the waveform shop light. Using low temperature fluorescent lights in conjunction with the waveform shop light does seem to help reduce the eyestrain. The large red and green phosphor peaks may help counteract the large blue peak of the shop light.
This also seems to correlate with anecdotal reports of installing a bare LCD panel onto a window to use sunlight as the backlight. At times sunlight goes even higher in color temperature than the 6500k of the waveform shop light and ideal-lume bias light. Yet it's very effective in alleviating eyestrain. This seems obvious since sunlight is completely flicker free, has perfect cri, and a smooth spectrum without any disproportionate spectral "peakiness". My other experiments with how various light sources interact with an LCD panel also correlate. Using an incandescent light bulb (100cri, smooth spectral, but mains flicker) as a backlight reduced eyestrain by a very large amount.
Another lower eyestrain display I managed to make was by modifying an old korean shimian qh-270 27" 1440p monitor. The backlight on it is interesting as it uses an alternating pattern of two different spectrum w-leds. One is a lower color temperature (maybe around 4000k) while the other is so high that it has a violet tint. Interestingly the high temperature leds were also the ones that flickered, so I disabled them by scratching the diodes out with a can opener. I then removed all but a single plastic diffuser layer behind the lcd panel. This modification reduced the eyestrain by a significant amount.
Another trend I've observed is the tendency for older CCFL LCD, plasma, and CRT displays to cause less eyestrain compared to modern led displays (which are predominately white-led) despite having flicker. This corresponds to their spectral outputs. The CCFL, plasma, and CRT spectrals are very "peaky" but the peak amplitudes are not as disproportionate to each other compared to the W-led spectrum of typical modern displays and/or the highest amplitude spectral peaks are in the lower energy green and red color phosphors. And in my experience of taking apart a number of more modern W-LED LCD monitors, the raw led color temperatures have all been high enough to have a somewhat blue to violet tint (almost certainly higher than 6500k). I do wonder if this is why modern W-OLED displays also seem to cause significant eyestrain for a lot of people here despite their myriad of advantages over LCD (not having the diffraction gating grid, much less flicker, better contrast, better antiglare/antireflective layers).
If this spectral peakiness effect is true, that means QLEDs and QD-OLEDs should be superior in terms of eyestrain than their W-LED/W-OLED counterparts, all things being equal. The ideal would probably be a full spectrum 99CRI LED backlit/OLED display but there aren't any out there at the moment (potential future modding plans? 🙂). I would like to know if anyone has had experience with this. As of now the Samsung S95b looks to be a promising, albeit expensive, display to test.
Here are some spectral graphs of some various lights as examples.
Ideal-Lume (original models, discontinued)
Waveform Lighting 6500k shop light
Various tablets
Various phones
CCFL LCD display
CRT display
Plasma display
LG OLED65E6V (W-OLED)
Samsung S95B (QD-OLED)
Samsung KS8000 (QLED)