Page 40 of 50

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 4:26 pm
by ROUBAL
I'm rendering a new image with the corrected scale and no alpha, and I can already see that the bottom of the jar is completely black ! :(

Edited :

I have tried to leave more space betwen the glass and the liquid without more succes.

I will try like in Indigo :in indigo you have to make the Glass and the liquid intersect a bit.

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 4:30 pm
by radiance
ROUBAL wrote:I'm rendering a new image with the corrected scale and no alpha, and I can already see that the bottom of the jar is completely black ! :(

I will try to leave more space betwen the glass and the liquid... With Indigo it is the opposite, you have to make the Glass and the liquid intersect a bit.

I have purchased indigo when it became commercial, but setting materials is so long that I almost never use it.
I'd say try again when we have MLT released, it will make a huge difference.

Radiance

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 4:41 pm
by ROUBAL
I'm currently modifying the liquid volume to try like in indigo, but with more intersection volume.

Edit : tested : Still black, like old burned coffee in the jar... I will wait for MLT...

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 5:07 pm
by livuxman
ROUBAL wrote:I'm currently modifying the liquid volume to try like in indigo, but with more intersection volume.

Edit : tested : Still black, like old burned coffee in the jar... I will wait for MLT...
Have you tried to put rayepsilon to 0 in kernel preview mesh? It solved me a similar problem in a scene.

Cheers

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 5:28 pm
by ROUBAL
@livuxman : RayEpsylon to 0 doesn't make much difference with 0.0001...

I have tried to increase the Raydepth to the maxi (1024) instead of 8 on my previous renders, and it is a bit better, but still very dark.

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 5:47 pm
by radiance
ROUBAL wrote:@livuxman : RayEpsylon to 0 doesn't make much difference with 0.0001...

I have tried to increase the Raydepth to the maxi (1024) instead of 8 on my previous renders, and it is a bit better, but still very dark.
the issue is that in order to arrive there you have a very very low probability.
with MLT, once a firefly is found, it will mutate it and it will explore the dark areas and they will converge much faster.

Radiance

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 6:24 pm
by mlody47
ROUBAL wrote:@mlody47 : :shock: You are right ! Hell, how di I missed that ? My watercolours box measures 6 meters ! :lol:

I can almost swim in the jar !

I will try to reimport a rescaled obj and render again.

Thank you very much !
Huge set You got there. :lol:

I wasnt thinking about black bottom , but better rendering speeds with proper scale.

Anyway I hope this will help with something :)

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 6:27 pm
by ROUBAL
It hasn't solved the problem, but at least I will pay more attention to the scale next time ! ;)

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 9:22 pm
by kampfisk
Hello Octane world!

Lately, I've been thinking about not contribute because you guys have gone to town with the previous posted pictures!
But I just had to give it a go aswell.
I'm a 3Dstudent since 6 months ago and I'm loving every second of it. I normally use mental ray render but the idea of a fully GPU based render sounded to awesome to ignore, it's great to finally spending most of the time fixing artistic adjustment (vignetting and gamma) instead of final gather samples -__-.

My contribution is one of the first serious models I've made. It's a CPU cooler called Thermaltake SpinQ.
The proportions and the engraved text is right down to the blueprint.
The mesh is pretty heavy, weighs in at 582826 triangles. Now afterwards I think I didn't have to have the text meshed, at least not on every fin =)
All the lightning was done using a HDRI-map. And the only post-production made was noise-reduction which was barely needed.

I used Octane 0.81 Alpha Build because the betabuild crashed on startup =/
Rendertime @ 32000samples was 6h and 20minutes using a GTX275

Hope you like it and I'm looking forward to the final product!

Re: Nvidia Fermi Octane Competition

Posted: Mon Apr 19, 2010 9:32 pm
by radiance
regarding the beta build crashing, what drivers did you install ?

Radiance