Search found 4415 matches

by Dade
Wed Jun 23, 2021 10:41 pm
Forum: User Support
Topic: Noise behind a piece of glass
Replies: 8
Views: 127

Re: Noise behind a piece of glass

What is the material of the bottom ground plane in this scenario? It's Matte material. It is an SDS path (Specular-Diffuse-Specular). It can not be rendered by a normal Path tracer or even BiDir: https://forums.luxcorerender.org/download/file.php?id=6075&mode=view See the caustic reflected on the m...
by Dade
Wed Jun 23, 2021 11:26 am
Forum: Development
Topic: OptiX intersection
Replies: 6
Views: 73

Re: OptiX intersection

Sure, you are welcome. Just try to pick a small topic to start with: it is a lot of code to look at if you consider the full spectrum of features (CPUs, GPUs, OpenCL, CUDA, Optix, etc.).
by Dade
Wed Jun 23, 2021 11:24 am
Forum: User Support
Topic: Noise behind a piece of glass
Replies: 8
Views: 127

Re: Noise behind a piece of glass

I loaded the .blend but I have sincerely only a vague idea of what I'm looking at or what is the expected result. Anyway, you are using a "Two side" material with glass on one side and "Null" on other: it can not work. You are building an asymmetrical transparent material material that can not work ...
by Dade
Wed Jun 23, 2021 10:57 am
Forum: Development
Topic: New OIDN 1.4 with new features and better detail preservation
Replies: 42
Views: 2645

Re: New OIDN 1.4 with new features and better detail preservation

In latest sources and with OIDN v1.4, I'm pre-filtering only the Albedo pass but looking at the images I'm starting to think that it may be worth pre-filtering the normal pass too. I will do some test.
by Dade
Wed Jun 23, 2021 10:53 am
Forum: Development
Topic: OptiX intersection
Replies: 6
Views: 73

Re: OptiX intersection

I just wonder how do you combine rays emission/intersection/shading/shadowing on GPU side? Is it something like Wavefront Path Tracing used in PBRT (https://research.nvidia.com/sites/default/files/pubs/2013-07_Megakernels-Considered-Harmful/laine2013hpg_paper.pdf) or Megakernel approach? Each GPU t...
by Dade
Wed Jun 23, 2021 9:19 am
Forum: Development
Topic: OptiX intersection
Replies: 6
Views: 73

Re: OptiX intersection

I just wonder which engines uses OptiX intersection acceleration? PATHOCL and TILEPATHOCL when using a LuxRays CUDA hardware device (https://github.com/LuxCoreRender/LuxCore/blob/2899e622811dcbf4dc6fafcf257a1f3f4f174bdd/src/luxrays/accelerators/optixaccelhw.cpp#L52) I've made a couple of breakpoint...
by Dade
Tue Jun 22, 2021 11:31 pm
Forum: Development
Topic: New OIDN 1.4 with new features and better detail preservation
Replies: 42
Views: 2645

Re: New OIDN 1.4 with new features and better detail preservation

Can you post a cube scene with that material ?
by Dade
Mon Jun 21, 2021 3:50 pm
Forum: Development
Topic: Scene texture maps resize policy
Replies: 42
Views: 2352

Re: Scene texture maps resize policy

I may have fixed your problem: it was something extremely rare to observe on my PC (1 out of 20 tries) so I'm not sure. Can you try if the latest version, when it is ready (https://dev.azure.com/LuxCoreRender/Lux ... der/_build), works ?
by Dade
Mon Jun 21, 2021 3:17 pm
Forum: Development
Topic: Film-only out-of-core rendering support (CUDA-only)
Replies: 22
Views: 402

Re: Film-only out-of-core rendering support (CUDA-only)

AndreasResch wrote:
Mon Jun 21, 2021 2:35 pm
It should not matter if tiled or progressive rendering is used.
The topic of this thread is about the GPU memory used to store the film and you think it doesn't matter if you are allocating the GPU memory space for a full 7200x4800 pixels film or a tile of 64x64 pixels ?
by Dade
Mon Jun 21, 2021 2:23 pm
Forum: Development
Topic: Film-only out-of-core rendering support (CUDA-only)
Replies: 22
Views: 402

Re: Film-only out-of-core rendering support (CUDA-only)

AndreasResch wrote:
Mon Jun 21, 2021 11:33 am
I don't see what tiled rendering has to to with the capability to use shared memory or not.
One allocated a 7200x4800 while the other allocated a 64x64 film, I don't see how you can not see the difference.