2025-05-26 - Modeling module Meeting

Meetings
1

Day and Time: 2025-05-26T21:00:00Z

Attendees

  • Howard Trickey
  • Campbell Barton
  • Daniel Bystedt
  • Tariq Sulley

Development Updates
Campbell

  • Fairly happy with progress on UV synch select. Too many loose ends to try to get this into 4.5.
  • Question for Daniel: Dalai was pushing to remove the existing “not synchronize” option. Is there a reason to keep it? Daniel: very happy with the new stuff - checks all the boxes. Feeling is that it wouldn’t hurt to keep the old way around for a while. Would never go back to non-synch, but you never know what other people think. Campbell: the worry may be that we could never get rid of it if we don’t remove it now. Campbell: worried that there may be cases where want the non-synch. Would be happier if 3-5 artists said they would never use the old way again. Daniel: my feeling is leave it for a while and maybe in a year do some public polls to see if it should go.
  • Most of the other loose ends are details. But Python API needs attention.
  • Underlying changes to custom data caused some memory issues on high-poly meshes. Blender can freeze (4 million vertices). There’s a bug report and Campbell thinks he has a fix. Working on it.
  • Plans after UV Sync: maybe work on a UI project that hasn’t been announced. There’s also the tesselation project (like scanfill triangulation, but handles many more cases)

Howard:

  • Continued progress on bevel node, but now it is clear that the early June deadline for the 4.5 beta is out of reach, so this will have to be a 5.0 feature.
  • Looked at a bug report where Manifold Boolean messes up if there are flipped normals (actually some but not all are flipped in the example). Confirmed that this happens, but haven’t yet tested to see if this is what the library does or whether my code wrapping the library is at fault. Will continue to investigate. It is somewhat implicit in the definition of the boolean operation that the manifold mesh should define a volume of space, and that the normals point at the “outside of the volume of space”, so if the library doesn’t handle properly oriented normals then we may just add to the user documentation that the normals should be properly oriented.
  • While working on the bevel node, reimplemented the hole-filling algorithm using a modification of Catmull-Clark by Levin. Wondering if this is a generally useful function to be exposed as a library function and perhaps a Geometry Node or a Modeling Tool. Daniel: does sound useful – would love it as a tool. Campbell: grid fill does do a nice job in some cases but this would likely handle non-”rectangular” holes better. Howard: will probably package the hole filling code as a library function, and then possibly a tool. deadline, but still trying for that.

Next Meeting
2025-06-09T21:00:00Z.

Shared calendar of all Modeling Module meetings.

14 Likes
2

I tried UV selection as well and my initial reaction was that I was overwhelmed with selection options and felt like I needed a guide to understand all of that. I can imagine how it must be for the beginners. For that reason I would also advocate for removal of old non-sync selection method. Since it’s gonna happen for 5.0 we can take alpha and beta stages to collect cases where workflow regressed and address them before the release. Would be much better to have 1 very complete and easy workflow, rather than multiple ones, some being legacy.

6 Likes
3

Having taught some colleagues some blender stuff, I have not been a fan of their eyes glossing over in the UV editor for these exact reasons. I simply don’t think there is a use case for the way it was before. I would also advocate for removing the old method at 5.0

1 Like
4

The sync select build is SO GOOD! I’m loving it. Everything feels very intuitive.

Dalai was pushing to remove the existing “not synchronize” option. Is there a reason to keep it?

The important thing that the current un-synced mode does IMO is not selection related but actually visibility related.

If you have a model with multiple materials, and each material is unwrapped to its own UV space (common practice), having everything visible with sync select is very messy to work with.

image
image848×717 107 KB

Working this way, you have to hide and unhide parts of the mesh based on the material and it is all a lot of clicks (E.G. to switch from editing one material’s UVs to another: unhide, deselect all, switch to face select, select, select similar, hide). Turning sync select off was a decent way to manage visibility.

However, I think a better way to manage visibility is with an explicit visibility setting. My preference for the options to be:

  • All
  • Selected
  • Material Slot

But if it would be too much bloat to keep Selected around, then just an option for All or Material would be fine. If Material is chosen, it would just show the faces that are assigned to that material slot and there could be an active slot selector like there is in the Shader Editor.

image

But however it works out, I do think some option for managing visibility is needed so users don’t have to constantly select and hide parts of the mesh to make it usable in that common scenario.

10 Likes
5

A widget a bit closer to the grease pencil layers widget would probably be a better starting point than the material slot selector, but even a dead basic “one material at a time” option would be wonderful.

There are lots of situations in gaming where we’ll have a mix of tiling textures and decal atlases on a single asset, so anything that brings native material-awareness to the UV editor is a good thing.

2 Likes