How does colour clipping work?  Is it handled by POV or by image viewer?
The reason I ask is that I sometimes use high ambient objects for light
sources and if I use anything other than a pure colour (i.e. <1,1,1>,
<1,1,0>, etc...) the final colour gets clipped to a pure color.  For
example, if you have an orange light, rgb<1,0.5,0>, and make it really
bright, rgb<1,0.5,0>*10=<10,5,0>, the final colour gets clipped to <1,1,0>
or yellow. in every extreme cases, you could take a red declared as
<1,0.01,0.01>*100 and it would get clipped to white.  Is there anyway to
keep the colour clipping proportional to the original colour or is it
outside of the POV domain?

-tgq

Post a reply to this message

Trevor Quayle wrote:
> 
> How does colour clipping work?

It's the result of a limited dynamic range in 24 bit integer files.
If povray could write floating point files color clipping wouldn't
be necessary but then most of the image viewers would clip the
colors anyway. You would need to perform dymamic range compression
to be able view the high dynamic range floating point images
with standard monitor/DAC combination.

Povray can output 48 bit files with 256 times more dynamic range
than 24 bit files. For practical purposes it's as good as floating
point files but requires a bit tweaking with the scene file
to make the integer format perform like floating point format.
The idea is to control the clipping with a global scaling parameter
for light sources and do the compression in an image editor. It
works because light in a povray scene is a linear system. If 2x
light goes in, 2x light comes out. I got the idea after looking
at photorgaphic film response curves.

See http://www.pp.htv.fi/kkivisal/c/compress.htm


_____________
Kari Kivisalo

Post a reply to this message