Omniverse API for scientific post-processing and visualization?

[emmanuel.motheau]

Hi all,
I would like to create usual scientific visualization, for example isosurface, streamlines, etc…, in real time and based on data that I have in ‘X,Y,Z,var’ format and geometries in USD.

I wrote a post-processing pipeline in python to convert and manipulate data in VTK and export back to USD within Omniverse, it works in principle, but as one can imagine it’s not very efficient and user friendly. If I am correct, Modulus does the same thing in the FGPA demo, it is all about external data manipulation in VTK.

I was wondering if the Omniverse SDK provides functionalities to do such specific computations on data. Is there somewhere an API that would do something similar as VTK, but embedded within Omniverse and taking advantage of the GPU? In other terms, can I compute for example streamlines of a velocity vector directly within Omniverse without relying on an external procedure using VTK libraries?

Thanks for any information !

Emmanuel

[mati-nvidia]

Hi @emmanuel.motheau. Have you checked out the Modulus extension for Omniverse yet?
https://docs.omniverse.nvidia.com/prod_extensions/prod_extensions/ext_modulus.html

We also had a Getting Started livestream for Modulus recently: NVIDIA Omniverse Stream | Getting Started: Modulus - YouTube

[emmanuel.motheau]

Yes, but as I mentioned in my original post, correct me if I’m wrong, but I understand that Modulus is doing such post-processing with VTK, and probably through the HPCvis-VTK extension (that is only available on Linux).

I would like to know if Omniverse SDK can provide something built-in, or if going through VTK is the solution proposed by Omniverse? (which is fine to me, I like VTK, but I just want to be sure that I’m not missing something)

I was also thinking that I could write post-processing operations via Warp, like in a more classical “compute shader” approach, but this is probably reinventing the wheel?

Emmanuel

[mati-nvidia]

Hi Emmanuel. I just got word from the team. We don’t have anything tailored for that currently, but we are working on visualization capabilities more tightly integrated into Omniverse.