June 14, 2010, 17:31 (GMT) |
missed this from last merge |
June 14, 2010, 15:12 (GMT) |
June 14, 2010, 10:22 (GMT) |
Render Branch: sss optimization, to share irradiance cache between materials in preprocessing pass. This required fairly major changes, so that each tile renders all sss materials, instead of each material rendering all tiles. I tried further to do a single rasterization for all materials, but it changed the results too much, for example eyes would be missing shadow at the sides because occluded faces were included in the sss pass then and not contributing any shadow. Unfortunate because it helped performance quite a bit as exactly in these cases the irradiance caching needs to take more samples. |
June 13, 2010, 17:09 (GMT) |
June 11, 2010, 21:53 (GMT) |
Material IDs work with nodes now; the id defaults to the containing material's id, but you can also specify an id mask output node if you want. |
June 11, 2010, 12:40 (GMT) |
June 10, 2010, 16:06 (GMT) |
change lamp wkey menu for render branch (dont merge this back on its own) |
June 10, 2010, 16:00 (GMT) |
June 10, 2010, 08:03 (GMT) |
June 9, 2010, 09:08 (GMT) |
June 8, 2010, 07:47 (GMT) |
June 7, 2010, 16:22 (GMT) |
Render Branch: revert last optimization commit, gives problems with some nested node materials, will fix this properly as part of my next commit with some deeper changes. |
June 7, 2010, 15:44 (GMT) |
June 7, 2010, 12:11 (GMT) |
Render Branch: optimization to skip irradiance cache with node sss materials in some cases, should save about 25% rendertime on the baby dragon. |
June 5, 2010, 16:09 (GMT) |
June 5, 2010, 16:04 (GMT) |
Render Branch: temporarily revert commit 29079 from trunk with particle path changes. This is causing two problems (reported in tracker): * hair doesn't move when transforming e.g. an armature parent of the mesh emitting the hair. * particle duplis on hair (for feathers) are not rendering. |
June 5, 2010, 14:35 (GMT) |
Render Branch: improve least squares reconstruction for irradiance cache, in cases where it is underdetermined (few samples points). Hopefully solves flickering in some of the mountains that we have here. |
June 4, 2010, 15:41 (GMT) |
Render Branch: post-sim correction for hair bendiness, doesn't work very well in my tests, but perhaps better than nothing. |
June 4, 2010, 12:25 (GMT) |
June 4, 2010, 12:10 (GMT) |
|