Everything related to Maxwell Render and general stuff that doesn't fit in other categories.
User avatar
By Mihai
#393676
Please everyone provide more info when presenting tests. What GPU, CPU vs GPU time, and in case of materials, specify if they used any special features - for example this gold test, did it use IOR or a regular gold material?
User avatar
By seghier
#393677
Mihai wrote:Please everyone provide more info when presenting tests. What GPU, CPU vs GPU time, and in case of materials, specify if they used any special features - for example this gold test, did it use IOR or a regular gold material?
i shared the scene this better than write
and you can open the scene test and see the settings
other informations gpu cpu ... are in my signature
By burnin
#393681
:wink:

IMO, GPU doesn't matter for now as there's no support for:
- additive layers,
- SSS (subsurface scattering)
- SSSSS (single sheet subsurface scattering)
- Reflection
- Refraction
Until these are implemented, the product is IMO still not considered as production ready & am not wasting any time on testing and comparing results to GPU vs CPU.

Also, you can read text from the console:
WARNING: - Material ( fastfrosted ) has additive layers, which are still not supported in the GPU engine.
WARNING: - Material ( fastfrosted ) has subsurface scattering, which is still not supported in the GPU engine.
WARNING: - Material ( fastfrosted ) has thin subsurface scattering, which is still not supported in the GPU engine.
WARNING: - Reflection/Refraction passes are still not supported in the GPU engine and will be disabled for rendering.
And all these GPU tests show the same. This^.

About stats... if all are done on the same machine and user is experienced in conforming to academic standard, i see no issue in revealing just the final conclusion. Personally am taking all results with healthy dose of skepticism.
Finally no one is getting no credits, no discounts, no praise or award for their work done.

PS (offtopic note)
For now am considering about the proper physically accurate tests (no frosted glass dragon, but more intricate scene) on CPU & GPU with few other engines as well, since results on before mentioned link all look fake (resemble more to an artistic rough ghost glass).
Also IMO, from what i've learned & by analyzing the scene, believing that the specific test was to be representative of physically accurate rendering is just a bad joke:
- frosted glass material with non physically correct shader (until implementation of shading model capable of preserving energy, this is considered an artistic approach),
- only blue albedo backdrop (non-real),
- pure black environment (non-real),
- single plane emitter (non-real).

As is, am only planning and thinking about it... maybe, maybe. :roll:
Observing modern society, there's simply no motivation to do so... :lol:

Have fun and wish you all the best with your loved ones.
Keep loving all without prejudice
and maybe, maybe will become reality.
:D
User avatar
By dk2079
#393683
luis.hijarrubia wrote:Just to clarify, reflection and refraction are supported on gpu. Only reflection and refraction passes are not.
but there seems to be an obvious problem with self reflection on the gold dragon, or?

could you explain what we are observing?

Image
User avatar
By seghier
#393684
When i made tests i don't care about time because gpu in my laptop is very slow
And time is not important because that change between different pcs
The most important is the comparison between gpu and cpu to help developers and i don't use gpu for rendering.
Burnin . thanks but i don't have time to create gifs
I spent long time to working on a scene with modeling or retopology and fixing problems , setup , renderings .... Than preparing images
By luis.hijarrubia
#393685
dk2079 wrote:
luis.hijarrubia wrote:Just to clarify, reflection and refraction are supported on gpu. Only reflection and refraction passes are not.
but there seems to be an obvious problem with self reflection on the gold dragon, or?

could you explain what we are observing?

Image
I'm actually reporting it to look at it. It's not even self reflection as the ball is a different object and it's not reflecting the base.
User avatar
By Mihai
#393686
seghier wrote: i shared the scene this better than write
Right, but it will take you only 2 seconds to add that info in your post. This way everyone who wants to know won't be forced to download and check your scene just to find that out.
User avatar
By Mihai
#393687
burnin wrote: Until these are implemented, the product is IMO still not considered as production ready & am not wasting any time on testing and comparing results to GPU vs CPU.
Me neither. But if you did take the time to make a test already, and the time to post, it would make it more interesting to see a comparison with time vs the CPU version, if you did also happen to render a CPU version. The whole point of GPU is after all that it's supposed to be much faster than CPU in many cases. Otherwise we wouldn't waste any time on it at all.
User avatar
By Mihai
#393688
burnin wrote: For now am considering about the proper physically accurate tests (no frosted glass dragon, but more intricate scene) on CPU & GPU with few other engines as well, since results on before mentioned link all look fake (resemble more to an artistic rough ghost glass).
Inspiration:

http://www.rayflectar.com/p05-Misc/misc.html
User avatar
By seghier
#393689
Mihai wrote:
seghier wrote: i shared the scene this better than write
Right, but it will take you only 2 seconds to add that info in your post. This way everyone who wants to know won't be forced to download and check your scene just to find that out.
these tests not to help users or to let them know the time of my renders in my laptop
this to help developers and who want confirm that he can test the scene and always when i posted test developers asked or members of maxwell team asked for the scene
why you want know the time ? this is not important ; it is important in one case if i have a powerful gpu and cpu and i want compare speed
and there are many users who tested speed
User avatar
By lebulb
#393691
seghier wrote:thanks but i don't have time to create gifs
I understand Seghier, no problem.
(But making a gif with Photoshop takes 20 seconds :wink: )
  • 1
  • 6
  • 7
  • 8
  • 9
  • 10
render engines and Maxwell

well I don't think AI will remain like it is now. […]

Help with swimming pool water

Hi Andreas " I would say the above "fake[…]