I have this model with decent low-poly topology.
I have sculpted it up to 3 levels of subdivision with a multi-res modifier.
I have weight painted it pretty decently, so that the low poly model deforms quite nicely, but when I activate the multi-res, the articulations develop some weird creases and inexplicable deformations.
The only explanation I find for this is that I could have got too happy with the Grab tool in sculpt mode, and now the polys corresponding to each low-res poly are too far off and the whole thing doesn’t behave as expected.
Can this be it?
Any solution for this?
I thought trying something with remesh, or duplicating the low-res and high-res versions of the model and reassigning them to one another via a new multi-res applied to the low-res model.
Shrink wrap also crossed my mind at some point though I wouldn’t know exactly how.
Multires is meant for sculpting, not for animating. The whole idea is to bake texture maps out of the high density sculpt and use them to give your lower-polygon mesh the appearance of having greater detail while still being reasonable for animation.
No, don’t apply anything. That will make the high-res geometry real. All you want is the base mesh. You can use a subsurf modifier to smooth it after you get your textures baked. Don’t do anything that will make a bunch of unnecessary subdivisions. Do you know how to bake maps from multires?
just uv unwrap the low poly model. set the preview in the multires modifier to 1, then go to the render settings and at the bottom there are the bake options. select either normals or displacement from the drop down menu, check the bake from multires tickbox, be sure your in object mode and press bake. then save the image from the uv editor and apply it as a texture. either a normals texture or bump texture, depending on which option you chose
Well, the contours in my low poly model are different enough to the high res one to make it impossible to just apply a normal map and make it to look anything like the high res mesh.
At the very least I’d probably have to go with the level 1 mesh from the multi-res and maybe un-subdivide it for the weight painting?
That will animate terribly. You need to plan ahead better in the future. If your low polygon mesh is so different from your sculpt that the shapes don’t even match, you shouldn’t have used multires. You should have done the sculpt first with dynamic topology and then retopologized a low polygon mesh for animation afterward. Or you should have planned out your base mesh so that it would match the basic shapes of the multires sculpt you intended to do in order to get a good normals map. At this point, if you can’t get your base mesh to match your sculpted shape, you need to retopologize a new mesh for animation.
That’s not a problem.
I’m not working or anything.
I’m doing a personal project, mainly aimed at learning Blender, and learning is what this is.
I’ll see to retopologize a low poly version of my model.
Thanks for the reply.
Before you do, can you show what your model currently looks like? A wireframe of the base mesh with 0 levels of multires and then a shot of the full sculpt? Before you go off half-cocked and make a whole new model we should see if your base mesh is indeed unsuitable for this or not.
…within the limits of the base mesh’s topology. It won’t add any new detail, but if you got a bit to happy with the Grab brush for instance, it will stretch that area of the base mesh to roughly match the sculpted shape. Save before doing this so you can easily revert; sometimes it produces very stretchy ugly results. Apply base is a nice easy way to get long, skinny, non-planar faces where you once had a clean surface.
That’s one reason I wanted to see the difference between the two. If you know what to look for, it’s easy to spot places where this problem might occur.
From what I can tell, applying a normal map on the “unsubdivided” mesh wouldn’t do any good.
The first level of subdivision does yield a quite viable base for a normal map though.
This said, I’m not sure if I should continue to do skinning of the lowest level mesh.
When you bake a normal map, you should set the Sculpt subdivision level as high as you had it when you were sculpting, and the Preview value to 1. It will bake a normals texture based on the difference between these too. Then you get rid of the multires modifier (at this point I would save a duplicate in case you have to go back for some reason), and put on a subsurf modifier. The difference between the smoothed version (which I assume is 1 level of subdivision) and the higher detail is plenty for a good normals map. And your base mesh is fine for animation. Probably a few more edge loops than are necessary, especially vertically along the torso and legs, but not so many as to cause trouble. And your chest and shoulder topology are good too. I agree with Small Troll. Bake, rig, play.
So to make things clear, what would be the steps then?
Generate normal map for difference between level 1 and 3 (the subdivided meshes in the GIF I posted)
Rig+skin low poly
Once animated, apply subdivision+normal map
Got that right?
And yeah, since I’m trying to match this to source material, it’s a given that I’ll have to go back to sculpting when I get to see my model posed like the original, so I’ll see to keep my full detail mesh.
Sounds about right to me. Just a terminology quibble though: usually in Blender when you say “apply” it means to click the Apply button on the modifier or to do some other operation that makes a generated change real. So you don’t want to apply subdivision; just use the modifier.