Bad To The Bone (...or Bad AT Boning)

okay… I know the last thing this place needs is another boning topic, but I’m still learning and a little stumped - what tutorials I’ve seen are too basic or too complex (or incomplete/dead links)…

I’ve been asked to animate a short clip for someone’s movie which involves upper torso movement and lipsync.

At the moment, I’m using envelopes for simplicity. Would I need to add a seperate bone chain for the tongue (maybe switching to vertex groups - can envelopes and V.groups be mixed?) or do I have to manually create tongue movements with shape keys? The trouble here is the character has a ‘beak’ and long tongue, a separate bone chain seems more logical than Shape Keys for every state. Would an entire mouth rig be a good option?

Another problem I faced is the ‘pinching’ around the armpit. I’ve experimented with the envelope size and shape to get to a fairly acceptable result, but it still doesn’t look right to me. Should I be adding extra bones for the chest muscles, or would weight painting help…? I thought the idea of envelope range/weight was that there would be a gradual falloff between adjacent deformations, but I can’t see much change with tweaking those. So far, I can only get a barely acceptable look by moving/scaling the actual bone ends, not the falloff.

I appreciate there may be many ways to do this kind of thing - bottom line is I don’t expect particularly realistic results, nor do I want to spend overly long on setting up & producing this shot.

“Any help would be hot”…

Ya, a “boning” topic would probably get locked pretty quickly :smiley:

However, I’m always looking for more rigging / skinning information :slight_smile:

I don’t know why you think envelopes are simpler than vertex groups. My experience is just the opposite. I’ve found I’ve had to make the envelope size huge relative to the mesh group they are supposed to control, and therefore with many bones close together, it turns into a big mess.

… Or (probably) I just don’t understand how to use envelopes.

Vertex groups / weight painting seems pretty simple to me, and works.

Making a 3 or 4 bone chain / vertex group / weight painting shouldnt take you more than 10-15 minutes to setup for the tongue.

Armpit - you’ll probably have to add extra “support” bones / vertex / weight paint. Take a look at the Ludwig or Mancandy rigs (you can find links to them in my signature in the “Best of” thread).


heh, okay… I should have used the term ‘Rigging’.

I guess that would be just my prior experience. This is particurlarly a problem when using converted models from other software. You can spend ages box selecting vertices, but there will still be rogue ones hiding in the (near) exact position left floating, then you have to go back over the same regions over and over. Also, my understanding of envelopes was that the falloff did away with the need for weight painting (on simple shapes). I could be wrong on that one…

Well, I guess you’re a lot better/quicker at Blender than I am! One area of confusion in this regard is how to set up the constraints/modifiers such that I have the ability to flex the tongue, but the whole tongue moves with the head. I’m sure it’s all embarassingly simple to others, but at the moment I’ve tried parenting the teeth/tongue to the head bone… but when I flex that bone, the the head moves less than the teeth/tongue. maybe this is an artefact of envelope deformation…or maybe I’m just stoopid.

here’s my problem with those rigs - I did take a look at those before starting this setup, but the fact is that those characters are so darned skinny that pinching deformation is hardly an issue anyway! Maybe the solution is indeed within those rigs, it’s just not obvious to me. Of course, if I was to use vertex groups in the first place, the shoulder would be less likely to interfere with the chest muscles…

I loved the idea of envelopes when I first saw them, now I’m not so sure…

a “boning” topic would probably get locked pretty quickly

I guess someone had to say it. Glad it was you not me :wink:

The answer to most of the questions is “yes” - even when that answer means two different things are correct.

For a reasonably believable tongue in a moderately complex model, bones are the best option. Shape keys, driven or set manually, can also be used along with these bones. Shape keys on their own are linear which means that to get a tongue to curl believably would require more than one shape key.

I’m with Mike on envelopes. I think they’re probably fine for some things but to me the apparently “simple” solutions are often too simple to deliver the complex manipulations you might need. If you can’t get localised control over the deformation then envelopes aren’t going to give you want you want here.

As for the chest pinching, again the answer is “yes”. Extra bones and weight-painting could both help - as could shape keys. Change to vertex groups and try weight-painting first and see how you go. Next, try adding a driven shape key that causes the mesh to form specific shapes as the upper arm bone rotates (if it’s a short, one-off scene then don’t bother setting up a driver, just apply the shape key manually as you need it). Additional bones might offer a little extra control but you’d need to experiment with weighting.

I haven’t done much at all with that, maybe envelopes are better for that, I really don’t know.

I assume you mean if you want to create vertex groups manually, and not with weight painting? That’s how I was doing it before I learned about weight painting. (A feature that would be nice for weight painting … or maybe it’s there and I don’t know about it … is how to hide parts of the model. … Something that can be done in “manual vertex” / edit mode)

LOL, I doubt that. I may have exaggerated a bit :slight_smile:

With vertex deformation, that’s typically when more than one bone is influencing the mesh. Probably a similar problem with envelopes … more than one envelope influencing the mesh.


You’re just the forum’s ‘Yes Man’! :slight_smile:

Thanks for the advice and suggestions. I shall play with this setup some more and report back in the not-too-distant future…

Dang! That weight painting was such a nightmare…

In comparison, using envelopes (for the arms and fingers) was a breeze. Trouble is, I had to keep rotating the model over and over to get the poly’s I needed to paint - and don’t get me started on the problems of trying to weight paint the tongue within an open mouth!

Again, my suspicion is that the model/conversion resulted in a messy arrangement of poly’s (triangles as well as quads). So it not only produced erratic weight painting results, but the deformation wasn’t very smooth…

It did occur to me that if the original model was designed with the arms in a more natural position in the first place, the deformation issues would not be so problematic. I mean, who decided all characters’ rest positions should be modelled in the ‘Crucifix’ position? Do they think I’m going crucify Jarjar Binks? hmmm… not a bad idea, now I mention it… :wink:

However, all said and done, I’ve got it to a reasonable state prior to animating. But I’m still not sure how to animate the lipsync. I fear manually deforming the model to create shape keys would be difficult and time consuming, and I’m afraid I don’t even know what ‘driven’ shape keys are. At the moment I’m leaning towards the use of a face rig, and setting poses for face shapes. Is this advisable? opinions and help would be very appreciated…

I honestly haven’t seriously tried a face rig but I can’t see that it would make lip-sync easier or faster. I suspect the opposite - but like I said, I haven’t tried it so it’s purely my assumption.

On the other hand, if the mouth and cheeks are well made - ie: good loops - then shape keys are reasonably simple to make and a breeze to animate - and they are eternally editable if you find they’re not quite working out as you go along. Edit once and all keys for that shape are updated.

In a couple of hours you could easily make half a dozen mouth shapes and for simple animation (you said this wasnt to be “perfect”), 6 good shapes should get you some amazing lip-sync action. Another couple of hours should have 10 seconds of audio pretty much lip-synced. So, one evening’s work and you’re well on your way. You easily could spend that same time trying to add, constrain and weight-paint bones then still have to work out how on Earth you animate a full audio sequence with them (My assumption again, others may well disagree and I couldn’t argue against them).

Start by reading my lip-sync tutorial. It won’t show you how to make the shapes but seriously, if you can make the model, you can make the shapes.

The thing to understand about lip-sync is that 1: You don’t try to match every letter in a word - just the major sounds (many of which look identical anyway) and also 2: Lip shapes are actually not the most import part of associating a character with the audio - acting is far more important. So, a well animated muppet is believable even though his mouth basically just opens and shuts but a character with no head, body, face and eyes action will be unconvincing even if the lips are perfectly shapes for every phoneme.

On the issue of weight-painting internal geometry like the tongue, I can only suggest that one approach might be to assign all the tongue verts to one tongue bone via the Edit Mode select and assign process (if it’s made properly this should involve just a few loop selects and maybe a couple of vert selects to finish.) and then to drag that bone right out of the head in pose mode so you can re-assign verts by weight-painting to multiple tongue bones (assuming this was the problem). Put the bone back in place when you’re done.

One day I’m sure we’ll be able to hide geometry in weight-paint mode but for now you just have to be careful where you paint and where the over-spray is hitting. This usually involves a lot of screen rotation to isolate a region as best as possible.

I only just found about the alt-b key the other day from reading the

Noob to Pro/Cool Things page. Have you used it?

Hiding Mesh
In edit mode press Alt+b and select area you want to be displayed with LMB. Rest of mesh will be hiden, till you pres Alt+b.

Ryan also shows some examples of it in his tutorial : Making the inside of the mouth

It’s kind of funky to use, your suggestion about editing / assigning verts … moving the bone is probably easier, but this tool is also available.



Note at the beginning of that page it says :

“If one would like to animate relative Shape Keys in Blender, this unfortunately, cannot be done with Actions. (See Action Editor)”

That doesn’t seem to be quite true with later versions of 2.41 and current 2.42.

I’ve only recently discovered that Shape Keys CAN be controlled with the Action Editor (and optionally the NLA editor).

The key seems to be :

  • Setting the IPO type to Shape in the IPO window
  • Enabling the ‘Sets IPO to be included in action’ Icon in the IPO window (headless man).

I also noticed that sometimes the ‘Sets IPO …’ icon gets turned on automatically, (maybe from a previous invocation of it?). Also it’s not clear to me if the IPO type has to be “Shape”, as I seemed to get an Action name created even when the IPO type was Object … and the Action then worked (erratically / randomly ???) as an action, other times not… Setting the IPO type as Shape seems to reliably work though.


I only just found about the alt-b key the other day from reading the Noob to Pro/Cool Things page. Have you used it?

Nope. I seem to recall reading a question the other day about it (from you?) but I hadn’t seen an answer. I will surely investigate. Thanks.

Okay, had a quick play. Weird!!! But probably quite useful once I get my head around it.

Hi again, a little update…

I’ve been trying to get some shape keys done - which is more difficult with a beak than a humanoid ‘slit’ mouth, especially considering this is a converted model (ie, not made with a linked duplicate half mesh)… However, after realising you can get away with only using 3 or 4 basic shapes, I managed to get the lipsync to a reasonable state.

I had a bit of a problem with intersection with the teeth, and at the moment I haven’t added movement to the tongue. One thing I noticed was that it’s sometimes difficult to judge the rendered shape because the edited (rest pose?) is affected by the armature (I can only presume) once in object mode, so I have to try and compensate. Another problem was that I tried edited existing shapes, but when coming out of edit mode I saw no difference to the original state. What am I doing wrong?

I’ll be adding some ‘acting’ to the character soon… and whilst head movements are simple enough to keyframe, how would I generate, say, hand gestures? Surely I’m not expected to keyframe each finger? Can I set poses for hands only, for example?

I’ve seen plenty of tutorials for walk cycles, but little else. Here’s hoping for some more tedium-blitzing shortcuts and tips from you guys! Thanks…

You’ve got nothing to worry about. That’s an excellent start. There’s a couple too many shapes on “Master” so his mouth looks like it says three or four syllables where there’s only two. Keep your shapes simple in the beginning - go only for the majors then come back when it’s all done and see what really needs more definition. So for “Master” you’d probably just need “M”, “A” and a “dull A/O” shape for the whole word so it looks like he’s saying “Ma-uh”. A little tongue movement between the vowel signs could probably suggest the “S” sound with no additional lip shape. Once you add head movement, the individual shapes are more difficult to see so you can get away with more than you might expect (I do major acting first, lip-sync second, refinement last). Also, do your lip-sync looking through the camera so you can see what the audience will see.

How you do hand movements depends on number of factors. How important are they? How long is this animation? Will the model be used for other animations? etc…

Either way I’d use bones, not shapes. Shapes don’t go around corners very well so curling a finger with shapes would be unconvincing unless the action was unimportant (not clearly seen) or very fast.

If the animation is fairly simple and you only want vague hand gestures to add a bit life to the animation, you might get by with just one set of bones controlling all fingers together. But if you want complex animation, with finger-pointing etc, then you’ll need to add bones for each finger segment. From here the choice is FK or IK and animate on-the-go or use driven Actions. Again, time and future use will help decide which way to go.

So far, I’ve stuck with FK fingers on my character (I don’t do much animating at all so that isn’t much of a guide). What I did was restrict rotation of each bone via the “num” panel locks - and yes, I keyframe each bone. I do like the “new” way of controlling fingers using IK targeted to a stretch bone though. There is a video of this from the recent Siggraph Conference (Calvin posted a similar setup recently too which you can download for reference). It would take very little effort to set it up and would probably save a lot of animating time. When I get a moment, I plan to add it to my guy.

If the hands are visible for just three seconds then you have to wonder whether it’s worth setting up complex actions you might never use since setting up the actions might initially involve the same keying process as you’d use if you just did it in the animation proper. In other words, the thing to consider is whether it would be faster just to do the animation or whether you’re better off spending time up-front setting up Actions and Constraints to save time animating later.

On the edited shape keys question, you have to remember that when you exit edit mode, the shape usually returns to the non-deformed or pre-keyed shape (it might be keyed to a different shape). The way to test it is to either re-enter mode and make sure the changes are still there or to set the key slider to 1.0 and make sure it deforms properly (this will set a keyframe so don’t forget to delete it if you don’t want it keyed there.)

Keep it up!

I don’t hear any sound, using VLC (Virtual Dub won’t even play the video)

Hmm, not sure what you mean by that.

Was that after moving the shape slider?

Andy and Bugman can probably offer more info, but I can tell you as a beginner animator, it takes a lot longer than you think to just pose a character, never mind animating ! :slight_smile: this scene took me about an hour just to pose the male character ! I’ve also read that pro animators only average 3-10 seconds animation a week ! :eek:


Sorry about that. It played okay on my PC at home, and my mac at work. I’ll try another format for the next WIP.

Simply put, when an armature has deformed an object, entering edit mode shows the vertices and mesh in the shape before the deformation. Similarly, when adding a shape key, the editable mesh in my file was not quite the same shape/position as seen in object mode. Maybe I needed to click the rest position button first, but I ended up trying to manually compensate for this deformation. All said and done, the current mouth shapes are acceptable for this short clip.

Before, after, during… :slight_smile:
To illustrate, say I felt my ‘ee’ mouth shape was too extreme and cartoony, and I wanted to tone it down. I’d select the shape key ‘ee’ and go into edit mode, manually adjust the vertices, then exit edit mode. But the rendered view is the same as prior to that edit.
I mean, if I’ve set a slider to 1, it should reflect 100% of that current mouthshape, surely? or do I have to ‘refresh’ the set keys somehow…?

I’m well aware of the time character animation takes - but it doesn’t hurt to check if there are any shortcuts you don’t know about. In fact, I do sometimes wonder why I agreed to animate this clip for some guy across the Atlantic I don’t even know… :frowning:

Click (enable) the third button in the Armature-Modifier panel (Enable modifier during edit) (OFf by default) and the armature deform will be active during editing

Without any “action keys” set, the only time I’m seeing something different from edit-object mode is if I have the slider set to less than 1.0 (or whatever max has been set). Edit always shows the “1.0(max)” deformation (AFAIK).

If you have an action key set, and that key is for less than the max slider setting, and the current frame is situated in that key or it’s “range of influence”, then you’ll see differences between (EDIT) (full shape deformation) and Object mode (current key setting).


I mean, if I’ve set a slider to 1, it should reflect 100% of that current mouthshape, surely?

Yes it should and usually does. That’s one of the things I like about shape keys, you can tweak them as you go. Doe sit still happen if you change frames, test animate etc? Are you certain no other shapes are influencing the final shape (shape keys compound on one another - which is also one of their primary benefits :slight_smile: )?

If this isn’t working for you then either you have some weird problem or you’re not adjusting what you think you’re adusting (easily happens with Blender). If you can’t find a solution you may need to post a blend with a full problem description.

Okay, another progress report… I’m reasonably satisfied with the lipsync for now, but the overall animation is giving me a real headache…

For starters, the deformation is a lot worse on the elbows than the shoulder ever was but even that I can tweak (mesh/weighting) later, time permitting.

Unfortunately, I fear I’ve really messed up the armature. For this short clip, I haven’t bothered to set up constraints, and am happy to keyframe by eye. However, for example, when I keyframe LocRot the endpoint of an ear and go back and forth in the timeline, that LocRot isn’t maintained. It’s as though something else is conflicting with it.

Additionally, the roots of the ears do move with the head, but they don’t rotate with the head (ie, honour their root positions in the back of the head as the head rotates). I thought I could get away with manually animating each necessary bone (as if it were FK) but I just ended up adding more Keyframe Konfusion to the mix… :mad:

I must confess, as many tutorials as I’ve read, I still don’t fully understand armature setup, including the difference between ‘joining’, ‘parenting’ and ‘connecting’…

I’m going to post a wip anim shortly and hopefully the blend file for any kind soul to troubleshoot…

WIP anim 2 (no sound)

Blend file.

Thanks in advance!

I’m downloading it now.

~5meg ! (Lowly dialup user here).

I don’t suppose you saved with the ‘Compress File’ option? :slight_smile: