> 1. when you use the new path tracing method, do you have the same speedups > when you turn on depth of field as you'd have with the REYES method? Cause > that's a bit of a problem with arnold atm.
Before replying to this question, one remark: in 3Delight, increasing the AA samples doesn't multiply the different GI samples. So for example: Scenario #1 - AA Samples: 4x4, GI Samples : 256 => Total resulting GI samples = 256 Scenario #2 - AA Samples: 8x8, Gi Samples : 256 => Total resulting GI samples = 256 (still) This means that you can increase your AA samples to get nice anti-aliasing/DOF/mb without the fear of an explosition of the GI samples (or specular samples). This has been done on purpose because we know that good DOF/MB takes a lot of samples. Having this in mind, you might find that 3Delight's path-tracer is very well suited to render DOF and MB effects. REYES would be unbeatable for extreme DOF with low to medium complexity geometry (I won't elaborate here unless specifically asked). The good news is that we have that too. > 2. And I guess I know the answer to that.. do you have the subpixel > displacement the same way as with REYES? ( my guess would be no to that, just > checking… ) The answer is yes. We do have sub pixel displacements in the path-tracer too. And there is no "pre-tessellation" so that memory won't explode either. -- Aghiles > > > On Wed, May 29, 2013 at 11:16 AM, Rob Wuijster <[email protected]> wrote: > video has been deleted? > Rob > > \/-------------\/----------------\/ > On 29-5-2013 4:16, Gregory Ducatel wrote: >> Hi, >> >> Sorry for the late reply... >> So since even in better resolution it is hard to read the setting panel, I >> have done an extra video: http://vimeo.com/67174304 just to show the >> settings for that scene. >> >> Regarding my configuration I am running a Windows 7 SP1 with 24 GB of memory >> and 12 Cores (2 x Intel Xeon E5645). Only 11 Cores were used. >> >> In the end, for shaders, 3DFS 4.0 is supposed to be release with it own >> material that can be used for "any" purposes but any way coding SL shader is >> definitely not that hard and you don't have to recompile every time the core >> is updated or when you want to do a modification of it... maybe since >> Aghiles is active on the list he can elaborate a little bit more on that >> side... >> >> Cheers, >> >> Greg >> No virus found in this message. >> Checked by AVG - www.avg.com >> Version: 2013.0.3343 / Virus Database: 3184/6363 - Release Date: 05/28/13 >> > >
