ANARI 1.0, VTK, USD, rendering improvements, physically based renders

I am looking for some rendered images, videos that were created in the last 1 year using VTK that show photoreal renders with features such as HDRI, image based lighting, ambient occlusion, ray tracing, global illumination, diffusion of light, physically accurate materials.

What am I trying to do?
I am trying to assess the work involved for creating a VTK based alternative to game engine architectural visualization solutions such as Unreal engine-Twinmotion, Unity, and film VFX render solutions such as Blender, Arnold, Renderman. I know VTK will handle very large mesh data and I can add more nodes if I run into large datasets. That level of parallelism is not available in commercial game engines or other CPU based renderers. What I am not sure about is the render quality I can get out of VTK, specially considering some of the algorithms for photorealism tend to involve a gather step that needs to see all the mesh data at the same time.

What are the recent developments that I know?
I see mention of openUSD (which is being adopted by the film, VFX and gaming community) in Khronos Group’s announcement on ANARAI 1.0.

More questions…

  • Can I expect the future integration of USD with VTK to happen through ANARI?
  • Should I see improvement of render quality in VTK since ANARI will connect VTK with Intel, AMD, Nvidia rendering engines?
  • Will these improvements come at the cost of reduced parallelism?
  • How far are we from importing a USD scene into Paraview?

Hi @abk

USD Support is being discussed here: OpenUSD Support

Anari support is being added here (by NVIDIA): https://gitlab.kitware.com/vtk/vtk/-/merge_requests/10324

Can I expect the future integration of USD with VTK to happen through ANARI?

No, USE will likely be integrated as an Importer, but Importer API we need to be improved to support all USD features

Should I see improvement of render quality in VTK since ANARI will connect VTK with Intel, AMD, Nvidia rendering engines?

Anari is merely a rendering backend. Rendering improvements are being added all the time, eg PBR has been added four years ago.

Will these improvements come at the cost of reduced parallelism?

Rendering in VTK is usually not the costly part. Computing is and this is unrelated.

How far are we from importing a USD scene into Paraview?

Supporting Importers in ParaView is a completely different issue. ParaView handle its scene and it would be quite a feat to take it away from ParaView. One could extract the geometry from a USD scene and show it to ParaView, this would not be hard to do. glTF is already supported.