POV-Ray : Newsgroups : povray.general : better radiosity in animation? : better radiosity in animation? Server Time
2 Aug 2024 20:17:44 EDT (-0400)
  better radiosity in animation?  
From: Rafal 'Raf256' Maj
Date: 1 Aug 2004 14:37:21
Message: <Xns9538D12C487E6raf256com@203.29.75.35>
Using radiosity in animation is realy a CPU-killer, because animation needs 
about x2 to x4 _better_ quality then still image, to get rid of annoying 
flickering.

Would it be a good idea maybe to interpolate radiosity colors between 
frames?

For example - low quality calculation informed that in frame #0 apple is 
lighten by rgb .5 from radiosity, and in frame #1 by .55 - it looks like a 
flickering bug, so we close-interpolate(*) thoes value and have result of 
about 0.51 with is simmilar to 0.50 and gives no distraction why watching 
animation.

In other situation - colors 0.5, then 0.9 - difference is too big to be 
just a bug probably, more likely - object moved out of shadow, or there is 
new light source, etc etc. Then we far-interpolate(**) values and have 
result of 0.85 for example - so that fast lighting changes that are NOT 
bugs are preserved (not motion-blurred).

(*) "close-interpolation(A,B)" as I called it will return value more 
simmilar to A.  interpolated = A*m + B*(1-m), where m=~0.1;
(**) same, but m=~0.9;

Implementation - currently POV stores  location of radiosity point and it's 
color (in some way, AFAIK). We add also frame-number to it, so we will 
store information frome current and 1 or 2 pastframes.

If current value is near to old value (it probably describes same object, 
or same place in scene space) it will be blended with old one, and too old 
values (mmore then 3 frames old) will be deleted to save memory.

In addition - if while calculating radiosti sample it looks like it is very 
simmilar to previous value, we can skip it's calculations and just assume 
that this point didn't change (and perhaps - calculare nearby points with 
lower radiosity settings). This should increase several times(!) render 
speed of almost-constant areas (in frames in with they do not change).

It should allow to render animations with even _lower_ radiosity settings 
and givind higher quality, especialy if scene is not too dynamic.

Any comments are welcomed.

-- 
http://www.raf256.com/3d/
Rafal Maj 'Raf256', home page - http://www.raf256.com/me/
Computer Graphics


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.