[Bf-cycles] Volumetric patch update

storm kartochka22 at yandex.ru
Mon Nov 26 06:08:11 CET 2012

Moderate changes, i think i solved all bugs related to MIS between 3
sample strategies, indirect/direct/light . Looking back, fixing 2 stupid
principal bugs take too long, but everything have end :).

I was depleted a lot trying again and again guess bug code lines by
staring at incorrect pictures, really need some time to recovering
before swithch to fix other things.
- transparent surfaces (play bad with transparent shadows, but it
absolutely needed for SSS -like scenes in bidirectional sampler), for
now they have double(or even more) weighted looking at pictures
- not symmetric adjoint BSDF for correct smooth shading normals for
light->camera rays, 
- DOF, for light-camera rays
- camera aperture direct hit from light ray,
- MIS between all subchains (checkbox named "Connect both paths" in GUI)
- spot light sampling, cameras other then perspective.

When all that will be fixed, maybe it can go some wider testing by brave
early adopters, on indoor scenes woth complex light, to see weak points
of that method.

And main mega bug that cost all other - it still can not run on my AMD
GPU, going to give it try to tweak it and run on Mesa with compiled
compute CLover, I think start it at Jan-Feb 2013 time, maybe earlier,
will see.

And thanks all blender ppls who keep helping with many fresh ideas,
crytics, that patch is not rocket science, as all described in many
papers 10 years ago and even more, but it never be written by me w/o you


I changed a bit root CMakefile to get it compiled on my Fedora 17, some
boost related error, feel free to revert it to svn state.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: blender_cycles_volume_mlt_spectral_52553.patch.bz2
Type: application/x-bzip
Size: 63424 bytes
Desc: not available
Url : http://lists.blender.org/pipermail/bf-cycles/attachments/20121126/f79060e5/attachment.bin 

More information about the Bf-cycles mailing list