Check the overview thread for more information about the meeting.
Present
- Sean Kim
- Hans Goudey
- Daniel Bystedt
- Julien Kaspar
Announcements
- Experimental Plane brush available for testing, see below for link.
Since the Last Meeting
- Sean - continuing focus on quality project, specifically test related topics.
Meeting Topics
- Experiment: Sculpt: Plane Brush & Devtalk feedback thread by @Nicola_Dessi
- Sean points out the feedback about the Plateau Brush
- Julian really likes the changes
- Moving in the direction of configuring brushes with brush asset settings rather than just changing to a builtin brush tool type is a good thing.
- Daniel plans on testing this too.
- Bend Boundary Cloth, Twist Boundary Cloth, Grab Cloth and Grab Random Cloth brushs don’t work with symmetry
- What’s the priority of this? Since it seems to have existed for a while this seems like not a high priority issue by default, but interested in further opinions.
- Julien agrees it’s not as important & not high priority right now.
- Hans says that this is a pretty fundamental limitation about the cloth simulation.
- Will lower priority / set as a known issue.
- Sculpting Layer Design
- Question of if it’s tied to a specific subdivision layer - Daniel says no.
- Hans brings up a concern of two separate things in the UI that are very similar (this + shape keys).
- Daniel brings up that tying sculpting layers to shape keys feels hazardous, due to how complicated they are.
- Thought experiement: What if sculpt layers were a modifier that could be anywhere in the modifier stack?
- Hans brings up a concern of this being tied specifically to multires, they seem like separate concerns & just mesh sculpting could potentially benefit from layers.
- This has implications related to the baked animation data (under the heading “Sculpting layers evaluated after pointcache” in the design task), as if it’s directly tied to multires then the workflow may be unintuitive.
- Sean asks about how we technically tie this to other data. If we have this directly connected to Multires it “solidifies” the design, that makes it difficult to undo in the future.
- Hans brings up the question about attribute support.
- This is also related to the fact that currently geometry nodes don’t really work with Multires
- There’s a lot of old design decisions that are catching up to us here.
- Hans brings up a technical design question of the relative “importance” this lends to the Multires workflow.
- Would be good to think about this in terms of 5~10 year plan / overall vision and avoid being tied to current implementation details.
- Julien brings up that the non-destructiveness and the deformations being stored in tangent space based on the base mesh are strong benefits of the multires modifier.
- Daniel asks if the design task should be more focused on the UI.
- Hans thinks this is unnecessary at this point.
- Julien asks if we would have to tackle attribute data on Multires first as a prerequisite for this design.
- Might be needed for mask + color attributes?
- Daniel mentions that from a user perspective that vertex painting on multires in the context of layers is way less important than just sculpting & tying these together doesn’t make sense as a first deliverable.
- Next steps
- How does this fit into the big picture long term design for sculpting?
- More technical design & options
- Sean to write up, exploring some of the other features and potential ways of designing this
- WIP: Add Geometry Nodes Evaluation to Sculpt Brushes
- Hans will spend some time seeing if the work started by @Nicola_Dessi can be ready for 4.4, possibly as an experimental feature
Need Help
4 Likes
I’ve also been thinking about sculpt layers outside multires, but can’t think of how it will work when geometry is constantly changed by remesh and dyntopo. How can layers be applied then? I can’t imagine practical workflow there, but I probably need to research this more.
I also feel like attribute support is good prerequisite, because without it layers design is weak. It would be especially good because it will explore how well multires works with geonodes and new mesh data, because that will be foundation of everything that needs to come in future for multires I think.
1 Like
I’ve also been thinking about sculpt layers outside multires, but can’t think of how it will work when geometry is constantly changed by remesh and dyntopo. How can layers be applied then? I can’t imagine practical workflow there, but I probably need to research this more.
The way most software handles this is to simply not allow topology modification when layers are used.
Yeah that’s most sensible, but in that case might as well use multires? 
If modifier is well optimized and doesn’t come with limitations like you can’t use attributes, doesn’t hurt to have layers actually interpolated over subd levels
I wouldn’t say vertex paint is “way less important” than just sculpting… I understand if vertex paint support is not a priority compared to fixing multires bugs and designing the sculpt layers system, but it is VERY important from a user POV.
2 Likes
Apologies - I think I mischaracterized this comment while keeping track of the discussion during the meeting; Daniel’s point wasn’t that coloring isn’t important for multires. The discussion was whether or not coloring in the context of layers was important & whether or not that needed to be a necessary feature of layered sculpting.
Edit: Just a note that I’ve slightly tweaked the wording in the initial post to reflect this.
1 Like
Oh ok, that makes more sense. Thanks for clarifying!
Though now that you mention it, it would be pretty handy/cool if you could save different color attributes per sculpt layer… I agree that would not need to be a priority, but such functionality would open up nice possibilities!
2 Likes
Yeah that’s most sensible, but in that case might as well use multires? If modifier is well optimized and doesn’t come with limitations like you can’t use attributes, doesn’t hurt to have layers actually interpolated over subd levels
This is essentially one of the main points of discussion that we had during the meeting - it comes down to a technical / architectural decision in the end, but the summary of the concern that Hans and I raised was that if layers are inherently tied to Multires, then it locks us into a specific design that makes support of other workflows difficult. It’s a tricky topic to navigate, and balancing scope without boxing ourselves in is why we have the summary of needing to do more investigation & technical design.
2 Likes
What other workflows are you talking about here? Something like shapekeys maybe?
Perhaps future shapekeys can be somehow based on geonodes (so they’re not locked to the base mesh data anymore and can happen at arbitrary points in the modifier list), then shapekeys on a multires object can essentially be layers.
1 Like
yes, sculpt layers should be done by shape keys and there should be a way to read the „hd“ details by subdivision.
Multi res for creating and reading.
Subdivision only for reading.
This would make things so much easier.
1 Like
This is admittedly an ambiguous answer - but what I mean is the general concept of doing sculpting work inside Blender. Whether that’s what we’ve discussed in the design task, dealing with shape keys, Alembic data, or something unforeseen at this point. If we rush into implementation and end up with a system that supports sculpt layers for only the Multires modifier, the specifics of how it’s coded could hamper future development or be enough of an effort to refactor that would make prioritizing or exploring similar work difficult.
1 Like
If layers are just VDM texture the maybe working first on how to implement layered texture painting would be a better idea ?
1 Like
Unfortunately, in my opinion, Texture Paint has other work that needs to be done before we would consider working on layered textures. I’m mainly thinking of the paint BVH refactor / 3D Brush work to have it share a similar backend as Sculpt, Vertex Paint & Weight Paint, though I’m certain that there’s more that I’m unaware of.
You do bring up a good point though about VDM textures and in general the inherent similarity of sculpting layers and texture paint layers. It’s something we’ll need to keep in mind with the further design work we have planned so that we also have more of a clear path forward with both projects.
2 Likes
I have a question. Is procedural texturing part of the layered textures project, or is it a separate project?
1 Like
Apologies - I’ve added some confusion here by being unspecific with terminology.
“Layered Textures” as a project and general Blender goal is something that’s better detailed in the original blog post - It goes well beyond anything that this module can work on in isolation. This is where the association with procedural layers comes in that you’re asking about.
“Layered texture painting” is a more general concept of how of sculpting layers could be applicable to areas that aren’t purely related to the multiresolution modifier.
In both cases, and as mentioned in the original blog post that I linked above, the general improvements to Texture Paint performance would be part of the necessary work.
2 Likes