[ATM] Wavelengths of light

[vladimir sacek]

Dominic-Luc Webb wrote:

> The human eye is a very nice match to sunlight. Somehow 525 or
> 531 nm is stuck in my head. The G in RGB and the peak of
> fluorescein fluorescence are all very close to this peak as I
> recall.

It is certainly a very close match. The sunlight energy flux actually
peaks at ~460nm (both, before and after the atmosphere), which is
all blue. However, the drop at ~550nm is small, only 5%, or so.
This is for a clear, sunny day; cloud absorption may affect it
somewhat.

It is fairly well known that eye spectral sensitivity changes with light
intensity, from ~550nm in bright light, to ~510nm in low light conditions.
What is less known is that eye contrast sensitivity also changes with both,
detail size on the retina (Campbell-Robson chart), and illumination level.
In bright light, the perceived contrast peaks at about 7-8 arc minutes
detail size (typical MTF bright/light bars form), while dropping nearly
five fold at ~1-degree detail size and to near-zero at the eye resolution
limit of ~1 arc minute (~60 cycles/degree).

As the illumination level decreases, both, peak sensitivity and cutoff
frequency shift toward larger detail size (graph:

http://webvision.med.utah.edu/imageswv/KallSpat24.jpg )
This could be useful for obtaining best possible contrast/resolution by
optimizing angular size of the detail on the retina (i.e. telescope 
magnification).

Vlad