Best reference for luminance values that are photopic (cone vision) versus scotopic (rod vision) is Donald Hood's chapter in Handbook of perception and performance. We have a good table of luminance, retinal illuminance, pupil size, etc. values from that chapter in the lab.
For even the dimmest bar (approximately .03 cd/m^2 using the 3% filter he made on top of the screen) in Alex White's buttonpress interception experiment, "scotopic" version, Dan, Alex, and I could see color. That is we could not make a bar created totally by the red gun indistinguishable from the grey bar that Alex used in the experiment. Actually for me, as I take the red gun and slowly make it brighter, it seems that as soon as I can detect the bar I can see that it is red. This suggests that the cones are close to the most sensitive mechanism in this condition, probably because these stimuli are parafovea. When I look at them in the periphery just several degrees (after dark adaptation) the stimuli all look much brighter and it's harder to discriminate the colors. This tells us if we want to do a truly scotopic version of Alex's experiment, we'd have to use stimuli that extend farther into the periphery. This probably also would overcome the issue of the bars getting so fuzzy and occasionally invisible when very dim- they were too close to the rod scotoma of the fovea.--Alex O. Holcombe 05:41, 13 June 2008 (UTC)