I would like to explain the issues that arise when using the displacement vray gpu within Maya.
The results when I worked with Cpu are as follows. (CPU.jpg)
As you can see, it can be seen that subdivision and displacement have been applied.
Now, if you proceed with the work using the GPU RTX render, it is as follows. (GPU_Issue.jpg)
As you can see, there is a phenomenon in which the displacement looks extremely lumped.
This error does not occur when there is no Subdivison option.
I want to know the reason why this problem occurs. And a solution.
In fact, internally, we solved this problem by applying OpenSubdivision, but I remember that OpenSubdivision was an option that was too heavy. I don’t think it’s a very efficient way. (GPU_OpenSubdivision.jpg)
Thank you.
Currently, Subdivision on GPU requires the “Cache Normals” option to function, therefore it is always active. If you activate it on the CPU, the results will be identical. We are working on bypassing this. For now, use the OpenSubdiv workaround.
thankyou.. open subdiv… too heavy… ㅠㅠ
As I continued to test, I found that pre-tesselling was applied within the vray to some extent, even if Subdivision was not applied.
Yes, this is normal and it is how Displacement works - the geometry is always tesselated prior to displacing. You can control this through the “Max subdivs” option in the Subdivision and Displacement quality attribute.
As you can see, the details of the GPU are significantly lower than that of the CPU.
GPU and CPU compute displacement differently, meaning a direct comparison may not be applicable.
Otherwise, we recommended lowering the Edge length or increasing the Max subdivs to bypass the “Cache normals” issue on GPU (in order to get similar to the CPU results). Another option may be to pre-subdivide the geometry rather than using OpenSubdiv.
Thank you so much. Thanks to this, many questions have been solved. After that, we will actually apply SUBdiviosn to the mesh to check the result. If I have more questions, can I ask you more questions?
