[Request] Cross-volume multimask

[jacobwhall]

Describe the feature or enhancement you're looking for
Currently, I believe MultiMasks can only be created on a per-volume basis. However, we have quite a few Sanborn volumes that coincide (published at the same time), and I think it'd make more sense to stitch them all together in one interface. This would make mosaics even cleaner, and make it easier to make fine adjustments to the multimask.

If this request is linked to a problem, please describe.
I think mosaic outputs could be even cleaner if we can seamlessly stitch multiple volumes together. Right now, it's difficult to go and find which volume to edit to adjust a specific spot in the mask, and the boundaries between volumes are even more difficult to work with.

[mradamcox]

You're right about the current capabilities, and the improvement you're describing here is something @rchampine I and have discussed in the past--it's something that would help the DSL work a lot as well, I just haven't been able to prioritize making it happen.

My working idea has been a new model called MapGroup, which I actually defined already, just haven't implemented at all.

A MapGroup would be the set of multiple volumes that make up an atlas, essentially a lightweight grouping of Maps. The idea is that if a Map is part of a MapGroup, then LayerSets of the same category (in practice, this would presumably always be "main-content" LayerSets, but could be "key maps" as well) could have all of their multimasks added to a single interface, essentially the same as the existing one, where adjacent masks could snapped together just as they already are.

The MapGroup is also a level of abstraction that will be needed for an upgraded city-wide viewer--modeling an aggregation of all volumes of a given edition as a single layer, even if they aren't actually combined into a single file as you guys are doing in Yesterdays.

Interested to hear other thoughts too. Ultimately, there are a few other things to improve within the MultiMask interface anyway, and I would probably combine that work with this.

[mradamcox]

There is a bit more about this structural concept here: https://www.ohmg.dev/concepts#multi-volume-atlas (and how it relates to other models within the setup).

[jacobwhall]

Yes, this MapGroup idea is exactly what I'm hoping for!

I'm wondering if there's a possibility of incremental development here. Even the ability to define MapGroups and browse them would be useful at first, and maybe that would make it easier to tack on a MultiMask feature later.

Maybe we even keep the MultiMasks for volumes, and produce those separately? I wouldn't mind treating them as separate products in of themselves

[mradamcox]

For sure, that's a good idea. Implementing a basic grouping mechanism can be the first step, later used to aggregate mosaics.

I wouldn't see changing anything about the multimasks, per se. They would remain an attribute of a LayerSet, i.e. an attribute of a grouped category of layers within a Map. But, the MapGroup would provide an indication of which multimasks to plop together into a single interface to start doing some snapping.