I've been trying to do some animations in the LOP context, using transform nodes. I'm beginning to wonder if this is the best way to do this. I create the transform node, and then I have to drag (or type) the paths for what I want to transform.

However, I haven't been able to figure out what the new node uses as the pivot point. It seems kind of like it is related to the pivot point of the last transform that I created, but other times it seems to be sensible (a good default would be the centroid of what I've given it as input).

Any advice on this? Should I use a SOP modify and animation the points inside that? What is generally the recommended way of animating larger, rigid objects?

Bumping this up since it never got a reply and I'm also curious about what the workflow around pivots in Solaris is. I can't seem to figure out how the Edit Lop pivot works.

USD has a specification for a "common xform API" which includes an explicit "pivot" transform which can be extracted and used for tools like the Edit or Transform LOP. But Solaris doesn't support either authoring or interpreting this common xform API (other than supporting the correct final equivalent xform). This is something that we keep wanting to do, but never find the time.

The Edit LOP's pivot point is just a setting of the handle. It is not stored on the prim anywhere. When you change the prim selection, we can either put the handle at the local origin or centroid (average point position) of the selected prim. You can then adjust the pivot point of the handle, but again, this information is not encoded on the prim anywhere, I believe.