OctaneRender® 1.024 beta 2.47 (lin/mac/win) [OBSOLETE]

Fri Jun 17, 2011 11:37 am

Have tested now the old 2.46b under Fedora Core 14 with Cuda Toolkit 3.0.
The issue that i can't close Octane, is the same.
Only kill -9 xxx helps...

face

kubo · Fri Jun 17, 2011 1:06 pm

First of all I know that this built is not "multi-gpu" ready, yet I wanted to test both gpus, my old 470 and my new 570 at the same time (I had a low power ups and I couldn't get both running at the same time till today). The thing is I know there are some issues with running on multi, but the funny thing is the difference on exposure or each one. I haven't tried the same shot with either one to see if it's consistent (I'll try later). Is this "expected"?

Fri Jun 17, 2011 4:42 pm

I got sometimes (not for every scene) a difference in treatment between right and left side of the rendered image, but it was not a difference of exposure : rather one updated faster than the other : right side more grainy if I remember well, and the difference is no more visible after a certain amount of samples per pixel. The problem had been fixed in previous versions with Cuda 3.0, but appears again with 2.47, so I think that it is a problem of multi GPU support in this version.

kubo · Tue Jun 21, 2011 9:52 am

in my little spare time I've done some tweaks to the model and materials, no straw this time, specs on the render. Caustics are fun! I hope to have some time by the end of the week and I'll post wips and upload the coke and foam material to the LDB

Tue Jun 21, 2011 12:51 pm

I have tried an interior scene with a number of artificial lights and daylight on both 2.46 and 2.47 . For some reason I had about 10 percent speed increase with 2.47 when using pathtracing.

But using pmc didn´t give the awaited improvement in quality. The parts of the scene with no or few daylight were still as noisy as with pathtracing. So apart from the new possibilities with caustics (which I haven´t tried yet), is pmc a real step forward?

Kind regards

stiwi

Tue Jun 21, 2011 2:56 pm

stiwi, this is a release candidate version released for early testing. PMC is still unoptimised and the current speeds are definitely not what the final version will have.

Tue Jun 21, 2011 4:45 pm

Thanks Daniel. I wasn´t worried about the speed so much. I just hoped for some improvement in noise reduction. But I went through radiance´s comments on page 2 of this thread again and now I understand, that this too is to be improved in the final release of 2.47.

Really looking forward to it!

Tue Jun 21, 2011 8:10 pm

stiwi wrote:I have tried an interior scene with a number of artificial lights and daylight on both 2.46 and 2.47 . For some reason I had about 10 percent speed increase with 2.47 when using pathtracing.

But using pmc didn´t give the awaited improvement in quality. The parts of the scene with no or few daylight were still as noisy as with pathtracing. So apart from the new possibilities with caustics (which I haven´t tried yet), is pmc a real step forward?

Kind regards

stiwi

Could you please post some examples/comparisons?

Thanks,
Marcus

Wed Jun 22, 2011 7:40 am

Sorry, I can´t I worked on those pictures for a client. But if i have some time the next days I´ll make up a scene for a comparison.

Fri Jun 24, 2011 11:26 am

abstrax wrote:
necko77 wrote:can some1 tell me, do i have to adjust glass transparney to let say 0.5 to get sun in interior
if i leave glass transparency to 1 i dont get sun in interior
coefficient of dispersion is set to 0.012

what is correct glass parametars now ?
Don't use dispersion for windows. Only use dispersion if you need to, because it introduces spectral noise which takes longer to clear.

Regarding sun through glass: I haven't tested it, but I don't think you can get sun light through specular materials. You still have to use the architectural glass trick. Maybe Roeland can correct me, if I'm wrong here.

Cheers,
Marcus

What is the architectural glass trick?