Iirc, many moons ago, you could use a pointcloud in a mantra shader with pcopen and pciterate to have the points influence your shader.

I do not know how to do this in Karma or MatX.
So take this situation for example: I have a large grid and on some part of it there are a bunch of points scattered in a nice pattern. I want to give the grid a shader where the location of the points will give the grid the nice pattern.

There are several cheats to do this, bake a texture, rasterize the points into a volume, but those are all resolution dependent, and I want to avoid generating large textures or large volumes if it can be avoided and the points can just be queried live in the shader, specially if the pointcloud changes over time.
Thought about using copernicus for this to import sop, geo to layer, but that requires a volume again (right?), and the layer will need to have a set resolution.
Other option, have the points attribute transfer some attributes to the points of the grid, but again: limitation of grid density.


Can this be done and if so how?

You look up points and their attributes from .bgeo/.pc files with Karma Point Cloud Read (kma_pcread) [www.sidefx.com]. Also works in XPU 🍪

AslakKS
You look up points and their attributes from .bgeo/.pc files with Karma Point Cloud Read (kma_pcread) [www.sidefx.com]. Also works in XPU 🍪

Thanks, I'll try that. Nice! I was typing "cloud" or "pc" but that did not bring anything up, so it's just point, no cloud now :-) This looks like exactly what I need, tnx a lot! Appreciate it.

Hey it's just me or it seems like the Karma Point CLoud Read node, even if it's working on xpu, it doesn't apply any falloff to the values? It's like with cpu render the falloff works fine, while switching to xpu breaks the falloff.

Hi,

I don't see any discrepancy in my tests between the two engines. It would be great if you could share a test scene (or better if you log a bug) that demonstrate the difference. Thank you!