I think I understand what you are saying but I think my initial explanation wasn't so good. Here is what I would like to be able to do.JimStar wrote:Imagine it as the "Movable proxy" attribute of instancer node has higher rank than the same attribute of meshes. So, if the mesh is included in instancer and instancer is being compiled, all meses included in it will be considered "movable" or not according to "Movable proxy" attribute of instancer. BUT. As soon as the same meshes will be compiled (in the same frame) separately from instancer - the "Movable proxy" attribute of mesh itself will be considered...
Something like this...
Let's imagine for now that particle instances don't exist.
I want to instance geometry A 40000 times and want it loaded onto the gpu as movable proxy and not part of the common mesh for memory saving reasons but I only want to load them on the first frame of an animation as they don't move.
I also want to load geometry B as a movable proxy as it has animation and I need it to transfer each frame.
At the moment I can:
1. set the animation mode to camera only Geometry A will only transfer on the first frame but my animated geometry B won't be updated.
2. set the animation mode to movable proxy and get both the animation on geometry B, and the memory savings for Geometry A, but Geometry A will also be re-loaded per frame (and with complex geo like trees, buildings etc. this can take a few minutes).
Of course particle instances do exist and I'd like this to work whether geometry A used an instancer node or not, or if Geomety A used an instancer node but there was also geometry C that was instanced without particles but also only needed to be loaded on the first frame.
I may of course be missing something in your explanation and maybe what you have in mind will work for this but I didn't read it that way.
T.
