<div dir="ltr"><div style><span style="font-family:arial,sans-serif;font-size:13px">Hi Storm,</span></div><span style="font-family:arial,sans-serif;font-size:13px"><div><span style="font-family:arial,sans-serif;font-size:13px"><br>
</span></div>> Bidirectional can do some cool images (posting now Jay Hardy kitchen </span><span style="font-family:arial,sans-serif;font-size:13px">example to testing therad on</span><br style="font-family:arial,sans-serif;font-size:13px">
> <a href="http://blenderartists.org/forum/showthread.php?216866-Cycles-tests-the-new-blender-CPU-GPU-renderer-of-awesomeness&p=2329482&viewfull=1#post2329482" target="_blank" style="font-family:arial,sans-serif;font-size:13px">http://blenderartists.org/forum/showthread.php?216866-Cycles-tests-the-new-blender-CPU-GPU-renderer-of-awesomeness&p=2329482&viewfull=1#post2329482</a><div>
<font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif">It's hard to tell from the two images whether this compare or surpass trunk quality/performance.</font></div><div><font face="arial, sans-serif">Can you post the same file rendered in the same machine but with trunk instead</font><span style="font-family:arial,sans-serif">?</span></div>
<div><span style="font-family:arial,sans-serif"><br></span></div><div style><span style="font-family:arial,sans-serif">It's nice to see progress in both fronts (volumetric and bi-di) regardless :)</span></div><div style>
<span style="font-family:arial,sans-serif"><br></span></div><div style><span style="font-family:arial,sans-serif">( btw, I love that kitchen scene, I use it for benchmarks as well </span><a href="http://goo.gl/DFy98">http://goo.gl/DFy98</a> )</div>
<div style><span style="font-family:arial,sans-serif"><br></span></div><div style><span style="font-family:arial,sans-serif">Thanks,</span></div><div style><span style="font-family:arial,sans-serif">Dalai</span></div><div>
<font face="arial, sans-serif"><br></font><div><span style="font-family:arial,sans-serif;font-size:13px"><br></span></div></div></div><div class="gmail_extra"><br clear="all"><div><a href="http://blendernetwork.org/member/dalai-felinto" target="_blank">blendernetwork.org/member/dalai-felinto</a><br>
<a href="http://www.dalaifelinto.com" target="_blank">www.dalaifelinto.com</a></div>
<br><br><div class="gmail_quote">2013/3/13 storm <span dir="ltr"><<a href="mailto:kartochka22@yandex.ru" target="_blank">kartochka22@yandex.ru</a>></span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Thanks.<br>
<br>
Yes, that was plan, but Blender is very big project, one man can barely<br>
understand every module, from probability theory to compiler graph<br>
theory and other inverse kinematics, graph black magic, so i deside do<br>
not burden core developers until patch pass some my own test and then<br>
wider testing by blender community, only then it can be passed to other<br>
developer attention.<br>
<br>
I am in very beginning state, patch do not pass my own testing, need fix<br>
3 bugs at least - final render crasher (so nobody will say "that your<br>
Blender is crap, it crash immediately as i hit F12 to render defaul<br>
cube, i go Maya instead and never return" when download new shiny<br>
ghaphicall build "now with bidirectional volume spectral sampling"),<br>
Adjoint BSDF (need duplicate and rewrite a bit almost every BSDF to<br>
respect difference between geometry normal and smooth normal) - lot of<br>
code, better postpone for late as it affect shaders system, that can be<br>
changed in core Cycles for some reason, to not do work twice, and ray<br>
visibility flags, as many scenes have hidden emission plains and other<br>
tricks.<br>
<br>
Only then call for some test build on graphicall, maybe with few example<br>
scenes to save ppl time changing endless debug checkboxes in integrator<br>
property panel. Then, after some time (~month?) if get community<br>
positive answer to question "is it needed at all, maybe better drop it<br>
and go only Path tracer way?" call for core Blender team support, maybe<br>
some sort of official branch. Ideally, it must work somehow on GPU,<br>
maybe in very cutted form, but main MIS paths must work and benchmarked.<br>
<br>
The problem is that bidirectoinal thing a bit complicated (although very<br>
vell described in Veach dissertation, and many pbrt-derived projects<br>
exist), i suspect that everyone that get how it work tend to run away to<br>
commercial dark side, and it is better to faster make base that can be<br>
later expanded by ppl with less knowledge to probability theory, but<br>
skilled in other related parts (like code optimization or good knowledge<br>
of GPU internals), but that base at least must produce absolutely<br>
correct image, and have all math algorithms in place. Now i have almost<br>
good picture, but terrific "obfscated" by leftower code, and unresolved<br>
adjoint bsdfs. I must solve it myself to save other time.<br>
<br>
In short, current state is volume things guarantee some work with bidir<br>
off, and only background used as only light. Anything other have issues.<br>
<br>
Bidirectional can do some cool images (posting now Jay Hardy kitchen<br>
example to testing therad on<br>
<a href="http://blenderartists.org/forum/showthread.php?216866-Cycles-tests-the-new-blender-CPU-GPU-renderer-of-awesomeness&p=2329482&viewfull=1#post2329482" target="_blank">http://blenderartists.org/forum/showthread.php?216866-Cycles-tests-the-new-blender-CPU-GPU-renderer-of-awesomeness&p=2329482&viewfull=1#post2329482</a>), but only one man can set it up and not crash - I am. If you very adventurous and lucky, you can make some street light in fog town scene (be ready for 2+ hours minimum to start to see some solhoettes of light beams).<br>
<br>
В Ср, 13/03/2013 в 21:32 -0400, Matthew Heimlich пишет:<br>
<div class="HOEnZb"><div class="h5">> Very cool stuff Storm, but without context it's kind of hard to follow<br>
> what is/isn't working in the current condition of the patch. Do you<br>
> think you could do another post to the mailing list with a simple list<br>
> like:<br>
><br>
> Volumes:<br>
><br>
> This works:<br>
> 1. ...<br>
> 2. ...<br>
> 3. ...<br>
> etc.<br>
><br>
> This doesn't work:<br>
> 1. ...<br>
> 2. ...<br>
> 3. ...<br>
> etc.<br>
><br>
> BiDirectional:<br>
><br>
> This works:<br>
><br>
> This doesn't work:<br>
><br>
><br>
> I think it would be very helpful for sorting out the status of your progression.<br>
><br>
> And congrats on crushing an old, annoying bug!<br>
><br>
> Thanks, keep up the good work,<br>
><br>
> Matt Heimlich<br>
><br>
> On Wed, Mar 13, 2013 at 9:10 PM, storm <<a href="mailto:kartochka22@yandex.ru">kartochka22@yandex.ru</a>> wrote:<br>
> > It Working !1!1111<br>
> ><br>
> > Fixed very stupid off-by-one array index offset in backward pdf<br>
> > calculation. I must admit, that bug tortured me more then 6 months if<br>
> > not longer. The problem it affect only paths with 2 and more bounces,<br>
> > and as energy usually lower it was barely noticeable on test scenes, and<br>
> > i have plenty of theories why sometime color bleeding too bright,<br>
> > blaming everything, from compiler optimizer up to FPU flaw. Now that<br>
> > nightmare over and my mind got a chance to stay clear.<br>
> ><br>
> > Many scenes converging so fast i cannot believe it possible (comparing<br>
> > to my old patch ofcourse, not other renderers) that bidirectional MIS is<br>
> > really cool stuff when it work (selected hard areas close to light<br>
> > covered by complex geometry).<br>
> ><br>
> > Also, i found that current algorithm that trying to track volime media<br>
> > material from light to camera direction is horrible, must to replace it<br>
> > by something different, so light sources inside object with volume media<br>
> > rendering wrong.<br>
> ><br>
> > Another bug is related to skin-like shaders, when "mix" or "add" used to<br>
> > combine other BSDF, then volume density get wrong values (i am think<br>
> > that random values used to mix clash with value used for distance or<br>
> > maybe other reason, but fact it completely broken), so only simple like<br>
> > single glass or transparent work.<br>
> ><br>
> > MLT replaced to more interactive approach, now you can set mutations to<br>
> > insane 25000 and get cool splashy useless pictures. But it force to<br>
> > single thread model as i do not want to hack deeper in Cycles and<br>
> > rewrite parts before kernel_integrate() call. I very doubt some one even<br>
> > test it, but must note that change.<br>
> ><br>
> > Other known bugs are same as previous patches, only preview window if<br>
> > light tracing enabled (segfault otherwise), only perspective camera w/o<br>
> > DoF, only flat triangles guarantee quality if bidirectional used, only<br>
> > zero size point and emission meshes as light sources (done sime initial<br>
> > support of spot lights, but it unfinished untested and buggy), no GPU,<br>
> > no OSL, no ray visibility flags respected for light->camera rays.<br>
> ><br>
> > Adjoint BSDF is next big target, as w/o it any surface with interpolated<br>
> > normals will have ugly errors, like flat with some weird gradients. In<br>
> > addition, I think that it is reason that bump not work for light->camera<br>
> > rays. No smooth surfaces, no bump maps -> unusable for wide testing.<br>
> ><br>
> > Short plan: get rid of stupid XXX_compensation constants and other debug<br>
> > leftowers, make code close to Blender coding style.<br>
> ><br>
> > _______________________________________________<br>
> > Bf-cycles mailing list<br>
> > <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
> > <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
> ><br>
> _______________________________________________<br>
> Bf-cycles mailing list<br>
> <a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
> <a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
<br>
<br>
_______________________________________________<br>
Bf-cycles mailing list<br>
<a href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a><br>
<a href="http://lists.blender.org/mailman/listinfo/bf-cycles" target="_blank">http://lists.blender.org/mailman/listinfo/bf-cycles</a><br>
</div></div></blockquote></div><br></div>