UE4 and Optix Engine to simulate Lidar

I am evaluating optix to simulate a Lidar sensor in the Unreal Game Engine, this is essentially a ray casting operation with zero bounces done a few million times a second in full 360 x +/- 30 degrees. Can anyone provide me any guidance on whether or not this is something that can be done (I didn’t see optix as one of the tools merged into UE4)? If it can be done, a pointer to some information on it. If it can’t be done and you can suggest an alternative, I would appreciate it. Thanks.

Hi. Now that raytracing has been officially announced for DirectX (“DXR”), that is likely to be a better fit for Unreal if you’re willing to wait a bit. See for example this forum:

https://forums.unrealengine.com/development-discussion/architectural-and-design-visualization/1471744-global-illumination-in-other-engines-unigine?p=1493653#post1493653

If you’re trying to measure raw performance, you could export meshes from UE and write your own Lidar simulator as a standalone OptiX program. It wouldn’t have to be too much more complex than optixMeshViewer. However, I do not have any experience integrating OptiX directly into UE as a tool; maybe someone else on here does.

Thanks for the response dlacewell. I appreciate it and like the idea for exporting the mesh and using OptiX outside UE4 but I think waiting for DXR is probably the better way to go. Thanks for suggesting that as well. Do you have any idea if it will simple be to call DXR to do a set of raytraces with no bounces to be performed on the GPU? Also, we’re looking at video cards, the GTX 2080 Ti is meant for DXR but the Titan V has quite a bit more power and faster RAM so I think the Titan is the better way to go. Any suggestions?

“Simple” is a relative notion. If you’re fluent with DirectX12 and HLSL, it should be rather straightforward to write a DXR application which just shoots primary rays. I would expect there is an example inside the DirextX12 DXR SDK which shows how to do that.

Since the Turing GPU architecture has hardware support in form of the RT Cores which will dramatically accelerate the BVH traversal and triangle intersection testing, it will be factors faster than any previous NVIDIA GPU generation for that task.

If you’re concerned about the VRAM configurations, there will also be the Quadro RTX 5000, RTX 6000 and RTX 8000 boards, as presented on the last SIGGRAPH, which all offer bigger VRAM configurations.