POV-Ray : Newsgroups : povray.binaries.animations : radiosity in animation-- a partial solution to 'flicker' : radiosity in animation-- a partial solution to 'flicker' Server Time
20 Oct 2021 15:29:07 EDT (-0400)
  radiosity in animation-- a partial solution to 'flicker'  
From: Kenneth
Date: 24 Nov 2020 14:15:03
Message: <web.5fbd5a8822e72c9d98418910@news.povray.org>
Radiosity in animation suffers from a kind of jitter or flicker problem, even
with high-quality settings... a result of the random nature of the rad 'light
patches' from frame to frame. Using a SAVED radiosity file, from an initial
camera viewpoint, would seem to solve that issue, but it's only partially
successful: As the camera moves-- and the scene's objects move as well-- the
saved radiosity file is no longer adequate to produce an expected smooth effect.

I came up with an interesting solution to part of this problem, at least for
POV-ray v3.8xx and maybe 3.7 too, which is more easily explained in an animation
test.

In my test scene, there are no actual light sources; the lighting is from pure
radiosity, from a skydome that has a cloudy-sky photo. I purposely chose BAD
radiosity quality settings to make the effect of the radiosity light patches
more obvious.

settings...
radiosity{
brightness 1.0
count 1
recursion_limit 1
error_bound 0.9
minimum_reuse .0009
maximum_reuse .0117
pretrace_start .08
pretrace_end .01
always_sample off
(all the others at their defaults)
}

The camera just moves from left to right...
location <0 + 5*clock,5,0>

Referring to the animation...
A) This is what typical radiosity light patches do in an animation, from frame
to frame-- completely random patch placement. Of course, this raw look would be
much less obvious with higher-quality rad settings, but it would still be there,
as a sort of minor noisy flicker in the animation.

B) This uses a saved radiosity file, made from a still image at the camera's
starting position. The animated scene's rad settings as kept as they are. As the
camera moves, there are a few extra rad samples created for the objects-- but
the initial 'unseen' parts of the scene require NEW patches, to fill in the
apparenty blank radiosity areas. They are like 'shadow areas' behind objects, as
seen from the initial camera position.

C) This is the same set-up as B, using the same saved rad file, but I changed
radiosity's pretrace_start and pretrace_end to 1.0 in the scene. This prevents
any 'new' rad samples from being created for most of the objects. It's an
improvement, but the 'rad shadows' are still being filled in with new patches.

D) This uses TWO initial saved radiosity files, one made at the camera's start
position and one made at the end (with different file names, of course!). Those
files can be opened and inspected in POV-ray-- they are just text/data files of
some kind-- then I simply copied-and-pasted the complete contents of the 2nd
file onto the end of the 1st file, which is then re-saved. This 'composite' rad
file is then used for the animation. In effect, the two different files, from
two vantage points, have overlapped and filled in the missing radiosity areas
seen in C). Again, pretrace_start and pretrace_end are both changed to 1.0 for
this final animation. There are still a few small areas that are not filled in
by this trick; they could be left as is-- or a *3rd* camera viewpoint could be
chosen to make a 3rd initial saved file...maybe from a vantage point above the
scene. The end result is smooth radiosity lighting during at least most of the
animation.

The only difficulty I see is when lots of objects move in the scene; perhaps
creating even more initial rad files could improve on that somewhat. But the
only way I can see to create a 'perfect' final rad file in such a case would be
to save a new file for every animation frame, then combine them *all* into one
final file! Not very efficient, of course; and the file would be HUGE.

For the entire time that I've been working with POV-ray, I assumed that a saved
radiosity file was a 'global' light map-- that is, creating a full 3-D map of
all areas of the scene, whether seen by the camera or not. That's not the case--
and I see now that such a map would usually be grossly inefficient and a waste
of RAM, for typical still-image renders which have only one camera viewpoint.
There is usually no need for light patches to 'fill in' the unseen parts of a
scene, such as behind opague objects...but animations could make use of such a
concept, if it was possible.

Here's a thought which I haven't tried yet: For a single saved rad file, I could
make all of the objects in the scene temporarily semi-transparent, so that at
least *some* light patches would be created everywhere in the scene. But at this
stage of my experimenting, I don't yet know how radiosity deals with
semi-transpartent objects.


Post a reply to this message


Attachments:
Download 'radiosity_in_animation_a_11_2020.mp4.mpg' (3382 KB)

Copyright 2003-2021 Persistence of Vision Raytracer Pty. Ltd.