I'm running large vellum sand sims, and I'm trying to render in Karma.
I've followed the common steps to instance sand pebble shapes on to each point, but the instancer in Solaris/Karma is taking 20+ minutes to cook before each frame will render.
Does anyone know faster instancing workflow in Solaris/Karma that can bypass this or speed up the process?
I'm just attaching a packed shape (16 variants) with an index attribute to 10+ million points
If it were Mantra, I'd consider something attributing the instancepath directly on the point for render, but I'm new to Solaris/Karma.

Any help would be great.

I'm seeing similar issues with the Instancer even with only 800 or so points to instance on. And a lot of crashes -- just switching from point cloud display to bounding box display or vice versa often crashes Houdini.

I can't even get the Instancer to work at all in Solaris right now; I got to the point where I have it showing me bounding boxes in the SOP context, but upon importing it into Solaris, I get massive CPU load and nothing to show for it.

There is a big difference between doing SOP import LOP with creating an instancer from packed prims and actually using Instancer LOP using just points and connecting static prototypes to it at LOP level

The former does a lot of unnecessary per frame conversion job so I'd recommend the latter even though it's not as convenient given that most workflows are already packed instanced primitives

How does that work with the biomes tools? Would I then create the points for the tree scatter and then feed that into an instancer in Solaris?

My first attempts using the instancer in Solares haven't produced any results so far. I proabbly have something not set up correctly.

Should I put the biomes scatter and instance network into the LOP instancer?

I had forgotten that you can build a node network inside that node, so I haven't tried doing that yet.