<div dir="ltr">Maybe you can take a look at Mitsuba Renderer, It has a good solution for solving specular transport case. <a href="http://www.cs.cornell.edu/projects/manifolds-sg12/">http://www.cs.cornell.edu/projects/manifolds-sg12/</a> </div>
<div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Feb 20, 2013 at 3:31 AM, storm <span dir="ltr"><<a href="mailto:kartochka22@yandex.ru" target="_blank">kartochka22@yandex.ru</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
As maybe some interested people can notice, recently i posting updates<br>
in longer intervals. The reason is i trying to solve perfect mirror and<br>
transparent surfaces weights. I was wrongly tink that it is trivial to<br>
fix (trying to focusing to only flat<br>
diffusion/glossy_with_non_zero_roughness), postpone that, and only<br>
recently notice that pinhole perspective camera have same issue, and my<br>
current code making wrong boost to light tracer samples (started from<br>
light), making final image worse then PT and even worse then w/o MIS at<br>
all.<br>
<br>
So, i really MUST to solve that, as even w/o any mirror /transparent<br>
material, any scene will have issues.<br>
<br>
The good news, is that when i make MIS weight offset manually (some ugly<br>
hackish black magic around debug_0 value ), result is somewhat<br>
comparable with what i seen Maxwell test scene, i mean not<br>
speed/quality, but overall feeling how noise converging near hard<br>
places.<br>
<br>
Currrently i trying to play with (L|D)S*DD*DE notation, better study how<br>
it related to path pdf, and other related things. Have some ideas how to<br>
skip unrelated bounce pdf from final, but far from real results.<br>
<br>
Damn Maxwell yellow juice material make me pissed off. (cannot resist<br>
and have download demo maxwell packet, trying basic "pyramid on floor"<br>
scenes) For a minute have strong feeling to fire disassembler to see how<br>
it do that ^^. But it is dark side way, and i doubt there is something<br>
unknown used, as arnold or many other, that is just good polished<br>
implenentation of well known algorithms.<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>
</blockquote></div><br></div>