!After conducting some tests, I'm not sure I correctly understand how karma calculates dicing on instantiated meshes.
The documentation page Karma User Guide Geometry [www.sidefx.com] indicates :
When dicing instances, Karma sets the dicing based on the nearest instance to the camera. Animated camera moves can thus sometimes cause flickering in the dicing, if the instances and camera distances are changing slightly between frames.
So I assumed it would either pick the nearest instance from the camera position or the nearest within the frustum as a base to calculate the dicing. But in my test scene I observed a weird behavior :
My scene is composed of :
- A bunch of instantiated long cubes with a material that uses displacement
- A camera located near one of the instances that rotate from bottom to top
As the camera rotates and more instances enter the frustum, the dicing resolution of the nearest instance decreases, even though the camera's position stayed the same (It only rotated in local space).
Between certain frames, the dicing drastically changes :
(I attached the full animation and the hip file to this post)
Do you think it's a bug or am I misunderstanding how karma detect which instance is the nearest ?
Additional details :
- XPU or CPU give the same result
- "Point Instancer" / "Instanceable Reference" modes on the Instancer LOP give the same result (As expected, in "Reference" mode there is no issue since each prim gets its own dicing)
