Thomas de Groot <tenDOTlnDOTretniATtoorgedDOTt> wrote:
> I understand from the documentation that this last one should always be 
> kept at 1 "for maximum realism" which is what I want.

  Well, the subject is quite complicated.

  The "maximum realism" comes from how lighting is calculated. When, for
example, light reflects from surfaces, the amount of reflection (which
depends, among other things, on the angle between the surface and the
incoming light) affects the amount of energy that reflects from the surface.
In other words, if the surface reflects 50% of the incoming light (due to
the surface properties and the angle of incidence), it will reflect 50% of
the incoming energy (measured in watts).

  This is different from what the human eye *perceives*. You see, 50% of
light (as measured in watts) does *not* look half as bright to the human
eye. This is because the visual perception of light is far from linear
with respect to the amount of watts that hit the eye. Instead, it looks
much brighter (the perceived brightness is more like 70% than 50% of the
original).

  On the other hand, display devices aren't linear with respect to brightness
either. This means that a pixel value of (128,128,128) does *not* send half
the watts as a pixel value of (255,255,255). The relation between pixel values
and the amount of watts that the display emits isn't linear either.

  Curiously (although I don't know if coincidentally), the brightness curve
of displays (at least those with a gamma of 2.2) is relatively close to the
perception curve of the human eye. This means that a pixel value of
(128,128,128) *looks* to the human eye approximately half as bright as a
pixel value of (255,255,255).

  That means that if you want "rgb 0.5" to *look* about 50% gray, then the
proper assumed_gamma is 2.2. This will cause povray to generate pixel values
of about (128,128,128) for that color (plus whatever modification lighting
causes to it, of course). Your monitor won't be emitting 50% of the watts
compared to pure white (which can be corroborated by comparing to an image
pattern that does, on average, send 50% of the watts), but your visual
perception of those pixels will be about 50% of that full white.

  The technical problem with using assumed_gamma 2.2 is that now surfaces
will reflect the wrong amount of light. (This is, AFAIK, a common problem
in all renderers, except perhaps the very accurate unbiased ones.)

  In most cases this isn't really a huge problem, though. The human eye is
very forgiving of such deviations from reality.

-- 
                                                          - Warp

Post a reply to this message