Hi guys,

I have more than 70000 instances that are not that hard to display and work in the Geometry context viewport, but when I use the SOP-Import LOPS to bring them into the Solaris, it becomes very very slow to load and display in the viewport.
I turned the Packed Primitives option to ON and tried all its possibilities, without success!

How can I fix it?



Thanks for helping.



Here is a simplified version of my scene:

The assemble sop in the network is re-packing the already packed geometry, which leads to nested packed prims where every top-level packed prim is unique. This causes lops to create 70K prototypes for the pointinstancer. I'm not sure what the assemble sop was doing because when bypassed the result appears to be the same (I might be missing something though) and the resulting pointinstancer in the lop network if much simpler, containing only a single prototype.

Here are four steps to improve performance when importing 70,000 instances into Solaris:

1. Use a Proxy Geometry : Replace instances with proxy geometry to enhance viewport performance.

2. Optimize Display Settings: Set the Display Mode to Bounding Boxes or Points to reduce viewport load.

3. Apply LOD (Level of Detail): Adjust Level of Detail settings to use simpler representations for complex geometry.

4. Switch to Houdini Engine Mode: Change Viewport Settings to Houdini Engine mode for better performance.

These steps should help improve the speed and efficiency of your viewport display.

I tried the Point-Instancer, but it's so slow, too.

Pointinstancer will be slow with 70K unique prototypes as there will be no instancing. The idea is to instance the same prototypes multiple times. Did you try bypassing the assemble sop in your setup? For me that avoids the double packing and makes things super fast.