Reom - Anatomical Mesh for MB-Lab

The topology note is important and I did figure that I’ll have to alter it after doing some tests with fat [which includes breasts], but that might take a while because it’ll break compatibility with many things so I’ll do it in one big patch with other things, until then I suppose I’ll solve it with proper displacement at higher subDs and re-arrangement at lower ones.

Very glad you pointed out the nipples issue :thinking: and your approach for it seems to be the most logical, I’ll have to experiment a bit but I’m very likely to end up doing just what you said

I’m glad you’re going on and on, that way I know what to prioritize, there’s no point in making the mesh if no one’s gonna use it.

And thanks, I’m trying to break some limits using displacement rather than shapekeys, shapekeys are additive and static, they don’t consider the normals direction, displacement kinda solves that problem and geometry nodes allow me to mix between them, still experimental but I hope I get where I want.

I wouldn’t mind getting the mesh to try out rigging. Would be interesting to see if I could modify it enough to get a convincing female form.

1 Like

This method of modeling is actually called quad spinning later called edge spinning/rotating in other apps. Some historical background, quad spinning was a well known and well practiced modeling method among some Lightwave users (those who were into organic modeling) including myself in the late 90s. This method evolved into “edge flow topology” later known by users of other applications.

It was called quad spinning because Lightwave originally did not have edges, so to spin an edge one had to select two adjoining faces and apply “spin quad” on the faces which would rotate the shared edge.

It was an excellent method for us back then (mid-late 90s) because

  • It was a super efficient method to create organic flowing forms using low polygon counts.

  • Lightwave actually had an initial version of smooth subdivision surfaces that worked relatively fast called metanurbs, which was a great way to visualize the efficient surfaces created by quad spinning method.

There are some dangers with this method however, and we can see them in the OP’s model. Too many or inefficient spins can lead to

  • star edges
  • floating vertices (a vertex sits on a single edge)
  • unrealistic forms
  • edge dips
  • pinched forms
  • bad forms during deformations etc

There is no reason to be hardcore on this method nowadays because the polygon resolutions improved and there are more advanced edge tools available in applications like Blender. The best is to use the combination of all of these methods as required by the visual goals.

3 Likes

That’s a pretty cool insight there, thank you!
I agree there are some downsides to my method depending on the usage, but at many times the pros outweigh the cons, differs from case to case as with everything

I dislike standard quad meshes for animation because of how their deformation works, especially in games where you have to set things up for real-time efficiency, high res solved things for single renders but it’s a hassle for animation, I’m trying to counter that with this method, and by combining it with new tech as you said

I’m open for any suggestions, we can all use some speed hacks if this works out now can’t we?

There are no hard rules when it comes to constructing visual materials, whatever works is generally a good approach as long as it is not too hacky. Too hacky is not good because it means that such asset will not hold up well in a real production. Experimenting and mixing different methods is definitely a legitimate way to come up with a better pipeline for these assets.

Obviously there are going to be some rules and limitations imposed by the production pipelines for things like rigging, animation and shading, but we have a lot of options to fix and rewire produced assets to better conform to those rules.

1 Like

I just noticed your comment, I sent out an earlier version here but I’ll try and send a new one ASAP, sorry for the long delay

If you ever get the time, I’d appreciate an in-depth feedback for the troublesome areas.
I’m keeping the mesh quad (but apparently I missed a few areas, will fix them ASAP) and some forms look “weird” in terms of the current standards but they’re meant to reproduce muscle blending during animation,

My main focus is deformation during animation because shape can easily be tackled once I set some key control points (the mesh is too low poly so every vertex has its place, or it won’t work right)

Are you up for it?

Sure, the main issue is that you focus on the techie side a bit too much. Basically while the edge construction and edge flows are all great ways to think about topology, in the end the forms need to look correct whether you have come up with the most mind-blowing edge strategy or not. In real life the form follows the function, but we do reverse-engineering in 3d CG therefor the function has to follow the form when it comes to these kinds of works.

1 Like