Which Red LEDs for sharp vision while preserving starlight vision
Starlight (Scotopic) vision allows us to see things at very low light levels, but it doesn't allow us to see any detail (or colour). I can read text, but only if it's 2cm high.
So if you need to be able to read normal size text or see fine detail, you need enough light to trigger the Cones used for Photopic Vision.
If you also want to keep your eyes ready to see at starlight levels, then you want to MINIMISE triggering of the Rods used for starlight vision, because it will take your eyes many minutes to build up night vision sensitivity after light has leached the Rhodopsin from the Rods. It's impossible to avoid triggering Rods while having enough light to trigger Cones.
To minimise leaching of Rhodopsin you want to -
- use as little light as possible
- use it as briefly as possible
- use light colour which favours Cones over Rods
Favouring the triggering of Cones over Rods is possible, because the spectral response of Cones (centred at 555nm) differs from that of Rods (centred at 504nm)
With LEDs we can pick fairly accurately the range of colours in our lights - the table below is a guide to which LED colour to pick.
The second column shows the RATIO of Cone sensitivity versus Rods - i.e the higher the number, the less you will trigger Rods and desensitise them, for the same level of Cone triggering.
The third column shows Rod sensitivity in Lumens per watt.
450nm 0.03 Blue
500nm 0.13 Cyan
555nm 1.00 Green
600nm 7.63 56.4 Yellow
620nm 20.8 12.5 Orange
640nm 46.8 2.55 Amber
660nm 78.2 0.53 Red
680nm 95.1 0.12
700nm 93.3 0.03 Deep Red
720nm 89.4 0.01
So it's clear that at the Blue end, Rods are actually more sensitive than Cones, but at the Red end the Cones are more sensitive than Rods.
So at 680nm, the same level of Cone visibility will cause one-ninety-fifth of the triggering of Rods, compared with 555nm.
I find 700nm LEDs still quite visible - although this could be due to 700nm LEDs emitting some energy at 680nm.
Re: Which Red LEDs for sharp vision while preserving starlight vision
Suspect there is inaccuracies in the numbers you present and certainly in your interpretation.
1) The cone response (under photopic conditions) peaks at 555, the rods at 505, so the chart of numbers you have presented are "normalized" to the photopic curve at 555 nm.
2) Your numbers are a comparison of Relative sensitivity of the two responses, NOT absolute sensitivity. Further, it appears from the numbers that this is a comparison of the curves near photopic levels.
3) The statements "So it's clear that at the Blue end, Rods are actually more sensitive than Cones, but at the Red end the Cones are more sensitive than Rods. So at 680nm, the same level of Cone visibility will cause one-ninety-fifth of the triggering of Rods, compared with 555nm." This is only slightly accurate at scotopic levels. At scotopic levels, rods are much more sensitive at the blue end. At scotopic levels, the cones are only marginally more sensitive than the rods deep in the red.
What your chart does show accurately is that from 650 to 700nm, the change is not huge in relative sensitivity. You can get some advantage by using a very deep RED led, but it is a matter of diminishing returns.