Just a closer look

Interesting. Visually it looks neat (I like that bar on the left hand-side). Nice job.

To understand correctly it's grouping the right hand side (Subsets) into 'randomly colored' groups per subset name so that still all the individual publishes of each asset can be seen. You are also unhiding an Asset column so there it's also clear from what asset each subset comes from.

I wonder how much speed-up this delivers as you work through it as an Artist. As I can imagine you'll still need to scavenge through the right-hand side. But I can imagine someone set dressing with solely models just selecting all assets on the left-hand side and and then limiting families to only model and solely browsing the right-hand side.

I'd love to try this out and maybe let some artists play to see if they have comments or remarks.

Also, I just wanted to link this video as a reference from Pixar on how they load setdress models. It could inspire more things.

To understand correctly it's grouping the right hand side (Subsets) into 'randomly colored' groups per subset name so that still all the individual publishes of each asset can be seen. You are also unhiding an Asset column so there it's also clear from what asset each subset comes from.

that's precisely what's going on.

I wonder how much speed-up this delivers as you work through it as an Artist. As I can imagine you'll still need to scavenge through the right-hand side.

To us a great speed up actually. We were just making a town and some forests. In both occasions we need to bring a lot of assets it and with this and consistent naming, it's a single import action. Selecting all assets that start with tree , all that start with bush (treeBirch', treeOak` etc) then right click on the modelMain group on the right side -> load abc -> done.

For trees for instance we'd have multiple variants as subsets in each asset, so Oak would have modelA modelB modelC and the same for other trees. If I'm bringing in 10 plant types and each has A,B,C variants that I can do it in one load instead of 30.

When we prepare the code to be compatible with core, we'd love for you to try it out and give more feedback of course.

Also, I just wanted to link this video as a reference from Pixar on how they load setdress models. It could inspire more things.

yeah that looks like USDs with proxy looks already applied. With avalon we'd normally first load models and and apply lookProxy to all. For rendering we'd change these to either standins or replace the looks with lookMain. Interesting thing is that they are technically importing the asset rather than subset in our terms, and it brings in the correct usd layer I assume. It's a bit different obviously because USD can include variants inside, whereas we only work with fairly rigid subsets.

Interesting thing is that they are technically importing the asset rather than subset in our terms, and it brings in the correct usd layer I assume.

This particular tool could just be forcing to load a "model subset" of course, hence it being simplified. However, I'll try and refrain from going into that particular video too much and try to keep this issue discussion related to your approach.

This feature could be interesting for making on-demand edits of different subsets.

This feature could be interesting for making on-demand edits of different subsets.

What kind of on-demand edits would you want to be making to different subsets? Not sure I understand your use case. :)

The workflow is to find the types of versions we want (animation, plate, lighting, compositing) in Ftrack, and RV (but any sequence player could do) to sequence them together for reviewing. With this feature we could do the same but in Avalon.

Just a note that this feature has been deployed to a few of our clients for a while now and the artist feedback is overwhelmingly positive. It's literally used on a daily basis, so we're very happy.

We'll prepare PR soon.