Help with horrible facial rig

hi everyone,

ok i tried facial rig, actually i tried to move the jaw of my character. but after i weight painted and moved the bone it was horrible! here r the pics. help. how to fix it. i have tried very hard to see if there are options to fix this, but only in vain. thanks

A number of factors affect how any mesh deforms, and it’s hard to tell just from the before & after pics what’s doing what. Some pics of your mesh topology, armature structure, and vertex weighting would be very helpful to analyze the problem.

At first glance, however, it looks like 1) your mesh is too high a density (too many faces) for the character’s shape, and 2) the vertex weighting is very uneven, which is probably made worse by the high mesh density.

thanks for your help n time. here are some more pics. no doubt i emphasized on modeling the head, which apparently messed up resulting in too many faces( if i didnt subdivide, the head wont be smooth. i’ve tried smoothening modifiers n stuff like that n learned that subdividing the mesh is indispensable for it, though i might be wrong), but i tried with simple meshes too( viz uv sphere with less number of vertices), but the horrible deformation still refused to get away. i’ve watched youtube tutorials( n many others) n followed the techniques. but mine results in ugly deformation while theirs results in “magically” beautiful synchronization. u understand right? i get so frustrated. thanks

PS: I m not an expert

chipmasque is right about one thing, that is a very dense mesh. That is actually the most dense mesh I think I have ever seen. There are 2 very easy ways to make a mesh look smooth. I’ve attached two pictures, the first one shows a character I’m working on, as is. It has 1037 vertices (that’s at the top of your blender window, to the right of the version number). Yours is at 63697 vertices. In the second picture, I’ve circled two tools you should use. #1 is subsurf. It sub divides surfaces, without actually doing so. It fakes sub dividing a mesh, so the mesh looks better without all the added vertices. Added vertices = more work for the computer. #2 is SetSmooth. It adjusts the normals (the direction light is reflected) to make the object look smooth. The second picture shows the mesh with both of those applied.

I’m sure chipmasque could provide you with better insights as to how set smooth and the subsurf modifier work, this is just a basic explanation.


Thanks revolt_randy for your help. I’ll definitely give it a try. but, can less number of vertices deform the mesh uniformly, unlike mine? i’m saying this bcz i’ve tried deformation with simpler meshes(with less no. of vertices) and moving the bone only results in the movement of ‘that part’ of the mesh which is weight painted. what i want is moving the ‘other part’ of the mesh so that it ‘blends’ with the part that actually moves. srry for sounding sophisticated.

Nope, you got the gist of it dead on!

Among the advantages of using subsurf are: The higher mesh resolution it provides is optionally used at render time only, making work in the UI more responsive, and (very important for weighting and animation), the much lower-resolution mesh cage of the model is not altered – so managing vertex weights and armature influence is hugely less complicated, because you’re only dealing with a handful of vertices per bone instead of bucket loads of 'em. When subsurf is applied during rendering, all the weighting of the lower-density mesh is automagically distributed evenly.

Basically the only reason to subdivide a mesh is when you need to add detail that can’t be produced with the existing mesh density – the general rule when animating is to keep your mesh as lean and clean as possible. That’s one reason methods like bump & normal maps were developed – they allow “faking” high detail on meshes that are actually fairly low density, using an “optical” effect rather than actual mesh shaping.

RE: SetSmooth – this is another “optical” effect used to make things look smoother even though they’re not. It averages the way light reflects across face boundaries (technically, it affects the face normals) so instead of a sharp division between faces (the SetSolid look), the surface seems to progress gradually from one face to another. But if you look at the silhouette of a low-mesh-density object which has been SetSmooth, you’ll still see the faceted look of the actual geometry. Game models show this very frequently if you look closely at them.

The amount of ‘blend’ between parts of the mesh when a bone is moved is controlled by weight painting of the bones. I think it would be easier to show you this with pictures. Attached are 3 pictures of my character. The first one is at rest. The second one has the mouth opened. (I did have the subsurf modifier on, but did not have the mesh to ‘setsmooth’, just so you know why it looks more refined than the previous pics)

In the second picture you will notice the bottom of the nose is ‘distorted’, it’s being pulled downwards when the mouth is opened. In the third picture, you can see why, the areas that are painted red move 100% with the bone. I intentionally did this because this a cartoon character of a comic book super hero. He’s wearing a mask, this is also why there is no details to the face, because a mask would cover them…think Spiderman.

In the third picture, parts of the mesh are painted red, some parts are yellow, some parts are green. Yellow parts only deform about 75% with the bone, green about 50% with the bone, cyan about 25% (there are not cyan parts in my pic) and blue doesn’t deform at all. This is how you ‘blend’ parts of the mesh controlled by one bone with the rest of the mesh. Also, circled in yellow, is the paint controls, where you can vary the weight of the paint, the size of the brush, how the paint is applied, etc…

Hope this helps,

thanks revolt for the details. they were very helpful. But i couldnt find the attachments u sent. also, i m new here, so they might be somewhere in this page that i dnt know.

thanks chipmasque for the additional details. another thing- is there a way to ‘unsubdivide’ a mesh? is there a way to model a mesh ‘in the edit mode’( like a human head)working on one side so that the other side simultaneously changes( kind of like a mirror reflection), so it reduces much work?

Given the extreme density of your current mesh, no. But look into the process of retopologizing (the Retopo tool), which allows you to retain much of your current modeling using a lower-density mesh. However, it might be a better learning process to just remodel the guy, since he’s fairly simple a shape.

The Mirror modifier is what you need to learn about and use.

Oooppss…forgot to attach the pics I referred to in my last post…here they are!!



Its not the rigging at all, its your mesh you need to learn to study topology at cg cookie .com and you’ll find many things to help you texture, with rigging, ipo etc there. the set up is completely wrong good try though.

Ah finally!! u guys are awesome!!! my problem is solved thanks so much…

But there is one last thing i’d like to ask. why doesnt hair grow on the part that is duplicated( by pressing alt-D)? i tried joining both the halves to make it one mesh. but after i grew hair on it, only the original half got hair. the other half was balder :o

Its weight group was not copied nor was the hair set up I have this problem when opening old files that I re import the hair is lost as it is not applied or converted to mesh and the groups with mirror does’t work depending on which version you use, I use 2.49c.

Just wondering:
Could a “Simplify density” modifier even be created… and if yes: could that find a way into 2.5 (or 2.6)?
It may happen that, for some reason, some wild subdivisions have been performed with no undos possible anymore…
Sorry, devs… i know you’re allready hard at work!! :o

hey Frank, how do you copy weight group and hair setup?