Cache only point transform for deformation

[Mzigaib]

I know this must be a really recurrent topic but I could not find anything on this so here is my question:

In SOPs if I have a deforming mesh with no point count change like a character or a water ripple is there a easy way to just save the mesh once and cache only the points positions in different frames like in USD? I thought Alembic could do that but I don't think it does.
I tried to do that with vex trying to store all the positions on a detail attribute but for heavy meshes it is really slow.

Any ideas on that? 

 

Thanks!

[tamagochy]

Use USD. All other formats cache full mesh per frame

[Mzigaib]

It is a shame,

Well thanks for the feedback

[haggi]

This is not correct. Alembic does only store the topoplogy once if you have a deformation without topology changes if you save as one large alembic file. If you need to use abc file sequences, it should be possible to layer alembics one with the topology and one with the changing points. Basically it's the same behaviour as in USD.

Edited May 3 by haggi

[Mzigaib]

Thanks for the response!
Yeah that is what I always thought and I know that with packed primitives you can do that only saving the transforms and applying it when you load it and there is a option for that but for deforming meshes with constant topology is really hard to check that and it is inconsistent with the file size sometimes it gets much bigger than "bgeos" so that is why I am not sure that is doing that sometimes.

Is there a way to check it or a option to enforce that?

Edited May 3 by Mzigaib

[haggi]

It is the case by default. You can simply test it by writing a sequence of some frames once as one abc files and then do the same as a sequence of single abc files. If it would save the mesh data every frame, the sizes should be roughly the same but the one single sequence abc file is much smaller than the single file sequence. If the files sizes in your case do not differ much you may have accidently changing topology... just did a simple test with a deforming torus and the single file sequence shows 18mb whereas the one large file has a size of 7mb.

Of course this depends a lot on the point data you have. If you have a some per point data, they will be saved per frame for every point so if there are a lot of them the difference will be quite small.

Edited May 3 by haggi