[Radiance-general] Luminance variation with distance

[Carsten Bauer]

Hi Charles !

I haven't examined the matter in full detail, but I think I may give a
first hint:

In rpict, an extra "source-drawing" function is called for tiny sources,
(resp. those appearing tiny) and this is done (most probably) for the
following reason: If an object is very small and far away, it may not be
hit by a view ray at all. That wouldn't matter much for normal stuff,
but for sources, this might be confusing. So this "drawsources" function
assures that at least a bright spot appears. 
I've tested it the other way, if the routine is disabled, a tiny source
becomes invisible when the distance is too big or the solid angle too
small, (ressembling the disappearing of your illuminated objects). I
noticed the effect already at solid angles of 1e-5 steradian, of course
it depends strongly on the absolute pixel resolution of the image. 

As said, I haven't examined this "drawsources" routine, but probably it
doesn't perform an exact calculation as its purpose is just to create an
impression of the source being there far away as tiny spot. This could
explain the deviations observed by you. Although, I am wondering about
this, too, as just displaying the original luminance would be no
problem. Hmm?!?

-Carsten