Hi.<br><br>Sadly I couldn't watch online the Cycles presentation on the Bconf. 2011, so maybe my questions were answered there... Anyway.<br><br>I consider myself a "neophyte" on this area, so excuse me if I say something stupid...<br>
<br>If I'm not mistaken, Cycles right now uses only Path Tracing to do the magic, which is easy to implement (especially on GPU) but really slow to converge on complex lighting scenes... I read somewhere that there are plans to improve the converce of Cycles by introducing Bidirectional Path Tracing + maybe MLT or ERPT implementations... Theoretically, this could improve by 2 or 3x the speed of Cycles...<br>
<br>My questions are:<br><br>1.- Can you elaborate on this plan?... Just out of curiosity I'd like to know your perspective about PMC (Population Monte Carlo), BiDir, MTL, ERPT and/or any papers you want to base on...<br>
<br>2.- Can you give a loose estimate on when can we start seeing this stuff?... I know this is an annoying question, but right now I'm starting to use Cycles for commercial projects and though I'm LOVING IT, I'd certainly benefit from BiDir and so on...<br>
<br>An example (45 minutes to render in a GTX 550 Ti, 4096 samples)... <a href="http://s5.postimage.org/9mjsycefr/Don_Melchor0001.jpg">http://s5.postimage.org/9mjsycefr/Don_Melchor0001.jpg</a><br><br>Thanks in advance.<br>
tuqueque.<br>