2022-11-15 Sculpt/Texture/Paint Module Meeting

Meetings
1

Practical Info

This is a weekly video chat meeting for planning and discussion of development related to the sculpting and painting in Blender. Any contributor (developer, UI/UX designer, writer, …) working on these features in Blender is welcome to join and add proposed items to the agenda.
Write us a message in the module chat if you’re interested.

For users and other interested parties, we ask to read the meeting notes instead so that the meeting can remain focused.

Attendees

  • Julien Kaspar
  • Joe Eagar
  • Daniel Bystedt
  • Ramil Roosileht

Agenda

Needs Commit

High Priority Bugs

New bugs:

In Review

See recent meeting notes for current patches.

New patches:

Current Topics

  • :gear: D16409 Fix inconsistent naming for cavity_from_mask operator. committed to master instead of 3.4 beta.
    Joe already back-ported it to 3.4 as well.

  • Joe will add an improvement of cavity auto-masking
    Instead of check-mark to use auto-mask in redo panel, pull settings from somewhere else.
    That would be more helpful. Will test soon for 3.5.

  • Release notes demo file

    • Julien will send the updated model to Ramil
    • The file could serve as a demo file for all auto-masking options
    • Add text for introducing all auto-masking modes.
    • Point to example tools for each auto-masking mode

Upcoming To Do’s

Joe:

Julien:

  • Wrapping up release notes
  • Brush asset Drafting blog post
  • Blender Studio work
  • Manual rewrite

Ramil:

Other Topics

Cherry picking

Brush Asset project team is still skeptical of focusing on :gear: D16320 Brush Properties Refactor right now. It was suggested to find a more generic approach, not just for sculpt mode but other data types and UI as well.

Useful for brush customization and right click context menu, and header tool settings.
But lowered in priority for now until cherry picking becomes important for brush assets.

Sculpt Mode boundary detection

Boundary tool is detecting boundary loops instead of the full boundary. The loop will always stop at corners.
Why is that the case and can we improve this?

Joe mentioned that stopping at corners is by design, yet the design reasons seem unclear.
There are some improvements in the sculpt-dev branch. He’ll look into porting them over.
Having boundary brush stop at borders could then be an optional brush setting.

Artists also pointed out that it might be too much effort for a feature that is rarely used so far.
We’ll need to discuss the priority of this again later.

Expand performance

Default geodesic behavior is initializing very slowly on high res meshes.
Even spherical expansion takes a long time to initialize.

If we want to keep using Expand for face set masking, we need to address performance issues!
Joe says it’s fairly simple to incrementally calculate geodesic distances.
All necessary refactors are now in master.

We should prioritize this optimization soon especially with Expand becoming more of an essential shortcut operator for creating/using masks and face sets.

3.5 Priorities for the short term

A discussion on what we want to prioritize in the short term for th Blender 3.5 release.

The immediate task is to finish and test :anchor: T98070 Improved Rake functionality. It’s a highly requested feature and already progressed far.

Auto-Masking also needs more improvements and wrap-up from after the 3.4 release.

Other Auto-Masking improvements are unclear in priority and could be pushed back for now:

  • :anchor: T102432 Sculpt Mode: Occlusion blurring steps + performance improvements?
    Joe thinks it’s doable. Like by improving the PBVH usage for the GPU.
    And reducing pen inputs for less performance impact from ray tracing.
    We also need to test this with Wacom on Windows. There it might be less of an issue.
  • :anchor: T101798 Mask from Cavity blurring artefacts
    Joe: Was waiting for implementing cotangent weights. Also necessary for fixing boundary brush behavior. Could be done both together.
    But it can unintentionally take week instead of days. Has to be seen.
    If it escalates Joe will give a heads up.

The current transform tools could also finally get some wrap-up:

  • :anchor: T80392 Sculpt Mode Transform Tool improvements
    The main task here is to recycle old patches for making the transform tools with tool settings fully functional.

    Ramil: Easier sculpt pivot transform is definitely useful.
    Daniel: Easier orientation is very important and needs a strong focus.

    There is still a need for a clear design on the pivot point UI/UX. Setting the pivot point via shortcut operator could be separate task then.

The biggest task that deserves attention is the planned Dyntopo refactor. Work on this can already start but could instead be targeted for 3.6. This would include:

  • Improved performance
  • Re-projected attributes while sculpting & preserved boundaries for attributes

Any attribute like UV maps, face sets, color attributes and even shape keys would then be supported for dyntopo sculpting. This should be a big focus.

This has priority over a multires refactor.
Multires will have to share some code with dyntopo refactor, so it makes sense to tackle them in this order.

Adding toolbar operations to the menus is less of a priority for now:

15 Likes
2

Are we talking here about multi-res support for shapekeys? :star_struck:

1 Like
3

No, it’s only about dyntopo support for shape keys.
Multires is a separate topic that needs to come after.

1 Like
4

Would the multires refactor also cover the serious precision issues it has at the moment, or is it focused around maintaining data like vert colors or face groups across levels?

mr2

2 Likes