I have objects created in Houdini and I want to share them between Houdini and Maya users so I am using alembic files which is working good so far, my question is:

- To maintain the best efficiency, space, performance, compatibility etc… should I pack objects before save as alembic or is it not needed ?
- Should I also convert to polygon soup before packing and saving or is also not needed?

Any tips or thoughts on this kind of workflow would be appreciated.

Thanks!

Anyone?

The way I am doing this now is after the geometry creation I pack and save as a alembic, if I use the copy SOP I pack and I also turn on the “pack before copy” option before saving.

I am just wondering if there is a better or a more correct way to save alembics.

Not sure about Maya, but its usually a good idea to use a path attribute to partition your geo. If you pack i think that is what takes care of the partitioning… added bonus, if you are using packed to do instancing, either packed disk or packed geo and you are saving to a single file the alembic will try to use alembic instancing where it writes the single instances to file once then just adds transforms for the frame range. I have had files go from 5Gigs down to 400 mb (where im instancing a lot).
I would try to use poly soups for most things. Ive heard that with only a few polys soups can be more inefficient.

I've been using packed primitives/Alembic workflow for my Houdini simulation->Maya workflow. It works very well but when I ran into an issue (tried to get the objects pivot points data) I realized it's all a bit black boxed. I understood through contact with support that what I wanted wasn't possible as the tech is still in it's early days.

That said, it's still way better than splitting geo up into different obj nodes. Exporting it all as geo cache O would loose transform data which isn't an option for me.

For static geometry I might go for unpacked polys through Alembic as that gives you a transparent preview of the geo. I've also worked with FBX but I wouldn't recommend it as I've spent a lot of time trying to track down data lossy data conversions. It's a bit stupid but best approach I had to read geo in Maya was an with Engine for Maya from where I loaded an HDA with a simple file node that I was pointing to a bgeo. This avoided conversion function on attributes that was happening in export/import for FBX.

- Should I also convert to polygon soup before packing and saving or is also not needed?

Yep.

Alembic has an equivalent data type to polysoups so yes definitely do this. Saves memory and more efficient to work with.

Thanks for the tips guys!