3
$\begingroup$

I'm trying to figure out individual steps of my ultimate goal from this question: Use Blender to get "Average" skull shape

And particularly, I want to duplicate what the authors of this paper did here:

Resampling of the mesh data was performed by raycasting based on a hemi-icosphere with 20641 vertices and 43284 faces. The directions of the rays were based on the vertices of the icosphere.

It sounds like, in order to create an 'average' skull, the authors made every skull have the same number of faces. This makes sense given what I know about shapekeys and morphing objects into one another. So, rather than tackle the entire project in one Stack-Exchange question, my question is very precise:

If I had a skull, or any object, with lets say 453,987 faces (random number) and I wanted it to be 43,284 faces (or some other specific number). Could I do that? Decimating or remeshing seem like they could get CLOSE, but is there a way to target a specific number?

$\endgroup$
3
$\begingroup$

You can try a Quadriflow Remesh (in the object Data Properties tab). Pick quad and select Quad and then hit the Remesh button:

Quad1

Before it performs the operation, it brings up an options window where you can select, among other things, the target number of faces:

Quad2

The results aren't always super pretty, and I only think I've seen once when it was able to remesh to the exact number of faces I targeted, but the algorithm generally selects what is best for the mesh (mathematically at least) and is usually pretty close to the target number. If the objects you are starting with (the different skulls) are relatively similar, they should produce quite similar results. At the very worst you will have to add or remove <30 faces manually.

Quad3

$\endgroup$
3
$\begingroup$

The shrinkwrap modifier projects a precisely defined source mesh (with desired face count, desired edge connectivity...) onto a target mesh.

It is not a guaranteed approach. A source mesh that is too coarse will not "sample" target detail. Imagine trying to capture Suzanne's ear with a simple plane projection. Nothing is automagick; you will have to somewhat design the source mesh so that it "sits more-or-less well" on the target. Target mesh before shrinkwrap Target mesh shrunkwrapped

$\endgroup$
4
  • $\begingroup$ I had already 'accepted' the other answer but thank you! I think this is an equally valid approach, I will try both and see what performs better. $\endgroup$ – Joe Crozier Feb 4 at 17:14
  • 1
    $\begingroup$ To go closer to matching the paper in question would use a hemi-icosphere (add ico sphere delete bottom half) and shrinkwrap using "Target Normal Project". $\endgroup$ – batFINGER Feb 4 at 17:34
  • $\begingroup$ @batFINGER I assume they just did that hemi-icosphere because of the part of the skull they were using in their paper (kinda the top half). Definitely easier to shrinkwrap I'm sure compared to the full skull I am trying to do with mandible and eye sockets and stuff. Will probably cause me plenty of problems that I've been brainstorming since seeing this. But at least I know where to start now $\endgroup$ – Joe Crozier Feb 4 at 17:45
  • 1
    $\begingroup$ @JoeCrozier only glanced over it, but suggest its about using a lattice of points that can be compared. Just having same amount of verts is somewhat useless if there is no order to them.. $\endgroup$ – batFINGER Feb 4 at 17:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.