Purpose:
There is a rigged human body.
My goal is to keep the skeletal bones in any pose.
These ‘skeletal bones’ are not rigs (Armature), they are objects that will be placed inside the body, even if they are not rigged to the body.
For example, the pelvic outline (landmark) should be preserved even if the body is deformed by changing the pose.
As if the outline of the clothes is preserved even if the clothes hanging on the hangers are pulled.
For this purpose, I’m looking for a way for the body-skeletal bones to interact like a clothes-hanger.
Cloth simulation
The body set as cloth will wrinkle and explode when simulated. This is because in Blender, seam UVs are separated when simulating.
Weight painting
After painting 100% along the landmark on the existing Armature weight, changing the pose, the mesh was severely broken. Also, since it’s subdivision 0, I didn’t even go into detail.
Shape keys
There are too many Shape Keys as they have to be created in all XYZ directions for all skeletal bones. It is also less accurate than the ‘body-skeletal-bone interaction’ approach.
Shrinkwrap (on “outside” mode.) It’s usually best to boolean join (probably via geometry nodes) all important features so that you can SW to a single object rather than doing it piecemeal, but this is expensive.
Dynamic mesh deformers. Bind, then edit the deforming mesh. Mesh that enters the deformer’s space will be deformed in a similar way to the deforming mesh’s deformation-- typically, deformed to push that mesh away. Can’t do a live boolean here since you need consistent topology, so order of mesh deform modifiers matters, and the mesh deformers often need to be temporarily scaled up in object mode in order for Blender to bind all vertices. Interpolation is better than with SW though.
Both of these require a lot of vertices to work well, and trying to do it with a high poly skeleton is going to be slow and tedious. My attempts at muscle systems have worked best with highly abstracted musculoskeletal meshes.
Shrink wrap using geometry nodes is the method I was looking for!
I’ve also tried shrink wrap before, but the results weren’t good, so I threw it away.
Why didn’t I think of applying geometry nodes?
Watching the tutorial now, I checked that it was applied more custom and dynamically through nodes. Very nice!
The Deformer method seems to be good too. It’s creative to use it this way.
Glad to hear it helped. If you’re having problems with the SW, I’d recommend think about using the “convex hull” feature of GN as well. When you have something like a high poly, thin pelvis, it’s easy for the skin to SW to the inside of the pelvis. By using a convex hull instead, you can avoid that problem.
Thanks for the tip!
Convex hulls seem to decrease the number of vertices. Can this number be maintained?
SW+GN vs Convex Hull - Which one is better for maintaining vertex count and topology?
(The vertex count and topology are very important to me.)
You can maintain it if you don’t do it live. Since bones are rigid, you shouldn’t need to do it live. If you do it live, the reduced vertex count is usually little benefit, since it’s made up for by the cost of computing the convex hull.
If you want to do this, you totally give up on meaningful topology. There’s no way around that. You just need huge numbers of vertices in your skin.
But also, it’s not vs. – convex hull is a useful part of SW.
