[thxraph]

I've been looking at the "Force Fresnel" settings, not sure why you need that for, + it is said in the manual that it is unapropriate for maps. I think I can pretty much have the same effect with Plastic coating in Arion, but doesn't seem like needed here.


EDIT: add Arion, 1h, CPU+GPU


Raph

[Aniki]

Hm, so the GPU doesn't help much in producing a cleaner image within Arion. I think in that case Maxwell would rather earn the tag "realtime" than Arion, don't you think ?

I was getting really excited when I saw the new feature list of Arion 1.6, the SSS looked superb, even more promising if its really much faster..

well, thanks for the efforts, thxraph.

cheers

Aniki

[thxraph]

First of all, Happy new year!!

Aniki, I don't know when i wrote something terribly wrong or explained myself in such a bad way that you misundertood me that strongly.
GPU does make the render significantly faster, and make it clean faster.
Later i'll post the dragon rendered with a more serious GPU so you can see the difference.
On my computer, the render is cleaner after 20min of CPU+GPU than after 1H of CPU only, and I have a low end GPU.

Raph

[Aniki]

First of all, Happy new year!!

Aniki, I don't know when i wrote something terribly wrong or explained myself in such a bad way that you misundertood me that strongly.


Raph

To you as well !

You explained everything very well and made that point with the GPU. I made my assumption based on the last two pictures you posted. And the difference between CPU and CPU+GPU after 1 hour is not significant compared to the same result Maxwell achieves in 1 hour.

No harm or disrespect intended. Its just how I see it.

[Half Life]

I never bought into the GPU thing (too many limitations), and I'm not a Maxwell user because of its astounding speed -- so the 1 hour comparisons mean very little to me.

I found the fully rendered comparisons much more interesting.

Best,
Jason.

[Aniki]

me neither, yet I found the faster response / approach to what the final image will look like interesting.

[thxraph]

Well, I can pretty much say that it my old GPU that doesn't help much. I also want to stress the fact that it is a bad case for a unidireccional renderer. If you render, let's say an interior archviz scene, or an exterior, the difference will be blowing. This scene isn't at all representative of a real production scene, as it uses a very poor lighting (1 small emiter) in order to intentiannaly put the renderer in a bad situation.

Here is what you get in 1h with 2 highends GPU, + a six core CPU. Maxwell's scene & Arion scene rendered on the same computer, ofcourse. (sorry for the picture being large)


Also, this remaining noise in the Maxwell render are the internal caustics, and technically speaking, it is way harder to get rid of this noise, or even integrates the caustics since the beginning than have a clean render fast. Clean is when everything is clean, don't you agree?

I'm pretty sure that it is what 2.6.1 intend to resolve, so i guess it's gonna be much better on the maxwell's side when it comes out. Then one will have to update the thread.

Raph

[Aniki]

now thats more like what Arion is promising ! thanks, thxraph for clearing (literally) this up

those dark dots on the Dragons cheek in Arion render will clean with more time, right ?

And was never my intention to put Arion in a bad situation, btw. I was very curious how it will perform with the new SSS.

Now I think I will have to save some money for new GPUs

[Half Life]

Looks like the roughness disparity was solved (was it a Force Fresnel thing?) -- they match very well now.

The dual GPU's definitely seem to make a difference for Arion, but obviously neither is clear yet and I would not use either as is.... and I suppose it is worth mentioning that Maxwell 2.61 will likely resolve some of the noise issues of the Maxwell version better so it may be worth revisiting in a few weeks.

Best,
Jason.

[Aniki]

means we might get Maxwell 2.61 in a few weeks ? any other info about improvements/features you can give away now ?

[numerobis]

yes, this strong caustic noise in maxwell is very annoying... and for me it seems to be stronger or more prominent in v2.6 (maybe 2.5 too) than in previous versions. Maybe because the overall noise level is better, i don't know, but it's very hard to get a clean image when these strong caustics appear - almost impossible for me in my recent projects...
maybe 2.6.1 will help here... (i hope so)

[Half Life]

means we might get Maxwell 2.61 in a few weeks ? any other info about improvements/features you can give away now ?

I'm just referring to what was said earlier in the thread:

I only suggest to wait until 2.6.1 coming in January because SSS is one of the areas that has been improved a lot in terms of speed - especially in difficult indirect lighting.

Just a teaser comparing 2.6 with 2.6.1, both rendered for 10 min:

Best,
Jason.

[numerobis]

i hope this improvement applies to "normal" refracted and reflected caustics too... not only sss

[Half Life]

Based on what was said I don't think it was a caustic optimization (although that may be part of it) but more about how the SSS resolves indirect lighting.

Best,
Jason.

[Aniki]

means we might get Maxwell 2.61 in a few weeks ? any other info about improvements/features you can give away now ?

I'm just referring to what was said earlier in the thread:

I only suggest to wait until 2.6.1 coming in January because SSS is one of the areas that has been improved a lot in terms of speed - especially in difficult indirect lighting.

Just a teaser comparing 2.6 with 2.6.1, both rendered for 10 min:

Best,
Jason.

ah, completely overlooked this, thanks Jason and Mihai