I just gave a try to use VDBs as velocity volumes in a DOP Gas Advect microsolver and it works like a charm, and I'm a happy person.
What I'm doing is I use a SOP Vector Field DOP to grab the velocity volumes (VDBs) from SOP and I'm advecting my particles with Gas Advect. My question is this (mostly to developers in this forum): are DOPs handle these VDBs as-is or do they convert these to regular Houdini volumes internally?
It will be automatically converted into a standard Houdini volume when it turns into a Vector Field. The DOP fields do not yet have support for the VDB data structure so always store it as a standard Houdini volume. However, since Houdini volumes are tiled and support constant tiles, this can still end up very memory efficient.
jlait It will be automatically converted into a standard Houdini volume when it turns into a Vector Field. The DOP fields do not yet have support for the VDB data structure so always store it as a standard Houdini volume. However, since Houdini volumes are tiled and support constant tiles, this can still end up very memory efficient.
Cool, I wasn't aware of that particular property of Houdini volumes (I'm not doing much HDK stuff.) So this means that even if the source VDB is very sparse, converting to regular volumes won't introduce too much memory overhead, I suppose…
Are there any plans of moving towards VDBs even further by directly supporting them in DOPs? (it might give performance improvements with large data sets)
