Using morph targets to open a mouth is terribly disappointing, chin gets smashed, upper lip crams into nose unrealistically, and I feel fairly sure that if the Jaw were wired as a joint, everything would look a lot better. Can this be done? Can we implement this in future models? Can we implement this in existing models? Thoughts appreciated -WTB

Can this be done?

It is done, but it requires a lot of extra work. In XSI we've occasionally set the jaw up as its own joint for "focus" figures, because it allows a range of motion that surpasses what's available through direct morfing. Equally, since most models stress the jaw-line already, the work required to figure out control falloff (are these present in Poser's rigging?) is minimal.

The downside is that you have to plan all of your morfs more carefully throughout the process, as a side-to-side jaw movement may result in a morf deforming badly. This is something to watch for with all cross-joint morfs, but particularly with the jaw, as the deformations of the lips are so closely tied to the movement of the bone underneath.

Additionally, a model that has a rigged jaw is much harder to customize. That's not a problem for high-end users who design and build their own figures for each use, but Poser users might get ticked if they found out that it took 4 hours to alter Vicky's facial structure to achieve the result they wanted.

Can we implement this in future models?

Depends on what "we" you're referring to.

DAZ might be able to do it, I'm not sure. There models tend not to focus on animation set-ups and rigging, so I'm skeptical about whether or not they could implement that change in a way that their customers would find beneficial. (In other words, doing it right might get them more sales, but doing it wrong would get them a lot fewer sales).

The other reason why not to do it is that it makes third-party development much more difficult and time consuming in some ways. Would merchants who make facial morf packs and characters be willing to jump through the big hoops that would be necessary?

Can we implement this in existing models?

I'm sure a lot of people in this community can, but they probably will find out very quickly why they shouldn't. Adding a bone to a pre-existing, pre-rigged mesh is an extremely time-consuming procedure. Adding the bones and the joints are actually the easy part.

The hard part is when you discover that you have to recast the entire figure's facial geometry to accomodate the new joint set-up.

If you really want this sort of thing in a figure, I highly suggest you pop into the big topology threads over at CGTalk and get a full view of the pros and cons, as well as what you'd be up against and what people have tried and had success with or failed with in the past. ;) (Note: These are just my thoughts, but please consider them as constructive. I am not a modeling goddess (yet), although I do have a lot of experience setting up and rigging in XSI, and understand most of the principles involved in mesh construction, even if I'm not yet good enough to implement them) Message edited on: 11/17/2004 01:10

Attached Link: http://www.poserfreebies.com

H.E.R. has a jaw bone and softriver is correct....when it came to 3rd party development a lot people had problems with it. If you would like to check HER out, she is free. --Rebekah--

How difficult it is depends on the model. It's easy to rig an animal model with long jaws, but hard to rig a creature with cheeks, such as a human. I sent a tutorial about it to Daz, but it's not up yet. But basically what you do is attach a bone to a few polygons behind the actual jaw, so the jaw uses that bone as a pivot point. The downside of this method is that it only works well with models that were made with the mouth open, because the jaw sometimes deforms oddly when opened further.

Attached Link: http://www.cane-toad.com/tuteRig_Facial.htm

You can do more than the jaw, if you want. In some cases, it may be useful to attach bones other facial attribtes as well. If you scan literature on general 3D character animation, you will probably stumble over one or the other tutorial showing you how to drive full facial expressions using bones, like the one behind the link.

Bloodsong added jaws to the p4lion and has used jaw boning setups in many of her own figures. its fiddly, but the realistic motion worth it. I've done some morphs for her dragonfactory dragon heads, which have articulated jaws, and its no worse than any other morph spanning 2 bodyparts

Lyrra, that's what I thought...why would it be any harder, right? -WTB

just out of curiousity, is the part that makes morphing the lips harder morphing both at once? i mean, if you had bottom lip morphs and top lip morphs, it should just be morphs applied to one body part and as easy as that usually is, right? again, i'm just curious...

When the jaw is controlled with a joint, the jaw is a seperate object, and has its own morphs that affect only the jaw, not the upper lip. (Unless, of course, you have a partial body morph that affects the top and bottom lips) It's not any harder really, you just have to make sure they lok right together.

actually, the way that we implemented the jaw rig in HER might be of some interest in this discussion: it was accomplished with some pretty exacting joint parameters; the geometry used for the group was actually the lower teeth. In doing it this way,we also included an upperTeeth group, effectively seperating the teeth from the head and the top teeth from the bottom, allowing morphs of the mouth area to be created without the concern of accidentally affecting the teeth geometry. Conversely, it also allows one to do extensive morphing of the teeth without the extra vertices from the head being involved at all. Another consideration is dummy bone rigging for the eyelids, as we have done with HIM and the Phibian. You have to be extremely careful with the joint parameters, but doing this can allow a much greater control of the facial elements in a way that is far more natural than morphs. In the eyelids its especially useful: morphing moves vertices in a straight vector; the outter surface of the eye is curved. Because of this morphs have to be sculpted to account for the protrusion from the curvature in many Poser figures (many of ours included in that...). This can cause blink, squint and any other morphs closing the eyelids to appear rather unnatural. By using a bone rig for the eyelids, one can successfully create a deformation of the eyelid area that curves based on the pivot point of the bone, effectively moving the vertices in an arc, naturally keeping the mesh "sliding" over the curved surface. We continuously experiment with methods of expanding what can be achieved with rigging in Poser. Methods of bone deformation for facial expressions are common in "high end" environments, but there is no reason the paradigms used there cannot be applied to Poser. It just takes some patience. :) -Les

Les-

That's the way we do it in XSI. Bones for the jaw and the eyes, occasionally another two or three bones for the eye-lids, if the animation requires it, connected along a dummy root to the head center, so they track properly.

Generally speaking, you only need to set the rig up once for each basic character "type" (male, female, child, dog, etc. ), then you keep the rig and attach it to your new geometry, unless the new character has special rigging that needs to be considered.

But, as per my earlier comments, if the rig isn't taken into consideration, then the mesh won't have the proper topology to implement blendshapes over the bone structure.

DAZ's characters weren't built this way, so you'd have to do a lot of extra work to make them look good with the new bone set-up, and many of your old morfs would be incompatible with your new rigging.

If you build a scratch character like this, it's not hard to do, and has definite advantages, but most Poser developers who specialize in character design and facial morfs/expression morfs might not like having to rethink their old habits (you'd have to ask them).

Additionally, you have to consider that most users will be resistant to the idea of mixing morf and pose techniques on facial elements. It doesn't seem like a big deal to us but the average user may feel differently, as they'd be forced to access stored poses or pose controls for general facial movements, and a separate set of morf controls for expression and musculature alterations.

Finally, any character made in this way would lose MIMIC support, which, if I understand correctly, is considered a big deal to animators. Would Poser animators be willing to trade the realism of arcing facial movements (which can be convincingly faked with morfs) for the convenience of Mimic, which saves them a ton of time? You'd have to ask them.

Again, I'm not trying to put a damper on the idea... it's one we implement every day, and believe is superior for a lot of uses. But I think that you need to evaluate the risk versus reward of doing this carefully. ( which is what this thread is about, right? ) :)

If anyone wants help doing it, we'll be glad to help them out as best we can. We're coming from an XSI background, and Poser's rigging is still pretty unnatural to us, but the principles cross-apply regardless of the application, and we have a ton of source material for rigging and animation that we can share.

Actually, HER, HIM and the Phibian are all totally Mimic compatible. The way to achieve this is by setting up ERC dials on the head object that control jaw and eye bone rotations at very specific degrees, and to make sure that the ERC implementation uses the targetgeom tags rather than the valueparm tags. In doing it this way, Poser "reads" the ERC dials as though they are morphs. The hardest part is just getting all the control ratios worked out so that the end result appears to users as though it's a morph when actually it's bone manipulations. So, there is no risk: we've worked it out so that it works just fine. HER is in the hands of thousands of users at this point and we have yet to recieve a complaint about the usability of this and I'd say with as many users as there with HER, one can make a safe bet she's seen her share of Mimic time. :)

-Les BTW: XSI background, meet Maya and Max background. Everything we impliment in Poser is based on methods we've used for the last nine years in one or both of those packages. ;)

Message edited on: 11/17/2004 23:41

And, in both backgrounds, the reason I've consistently said that there's a hell of a lot better ways to bone than people are using. I'm gonna have to dig out HER again and check out that rigging, Les. That's a pretty elegant solution tot he mimic issue. Applause for keeping it in mind. I've always hated the way we called them bones, too. It's not as accurate a term, it seems to me. But it's a hell of a lot easier to say than something like "motion control element", lol. Now if only DAZ would get creative in their rigging as well...

Add to the list of jaw-characters some of Lemurtek's P4 human-animal hybrids. They use the extra bone at the jaw hinge. Does the new DAZ Millenium Big Cat use any of these methods? Can their Millenium Cat yawn convincingly?

morph based. sigh...

well, from what I've heard a lot of people yawned convincingly at the cat... OUCH!! sorry... it was just too perfect an openning.... I'm going away now before Rebekah catches me posting and pulles my network connection!~!! -Les

Nice thread so far! Thank you Les and Softriver for sharing your tricks! And even yet, close to nothing has been done with scripting in Poser rigs: Not sure if it's a common practice with 3ds max (last time I tried, MaxScript was a PITA for scripting joints), but from what I gather, it's quite popular to use driven keys, MEL scripts or expressions in Maya rigs. There's Python in ProPack and P5, which would allow us to create much smarter rigs that do complex reshaping with a simple dial twist.

Les- I'm sorry if you thought I was trying to use my background as some sort of leverage for my thoughts. Quite the opposite. My intention was to explain that because I haven't been using Poser for very long, I still am new to the way rigging and morphing work. I don't doubt either your experience or your ability. I'm simply trying to evaluate the pros and cons of the initial poster's question.

Softdrive- I was just teasing. :) It's one of the big reasons I don't post much outside our own forums: people often times mistake some of the things I say in jest for seriousness, when about 90% of the time I speek with tongue planted firmly in cheek. I'm sort of a sarcastic bastard by nature so bare with me, please. :P For the record, though, I think there are a lot of misunderstandings with "high end" users regarding the nature of Poser rigging. While it does have some very distinct boundaries like the whole "only one parent to a child part" thing (look at a hand for instance: while there may be 15 bones in the fingers coming off the hand, they all have one distinct parent: the first joints of the fingers still have mesh from the hand between them) I find that there are lots of ways to "trick" the Poser system, you just have to be willing to do a lot of tinkering to get there. Stewer- It's true that in a lot of "high end" environments (I put that in quotes, always, because IMO it's a term of snobbery that has done nothing but hurt the perception the "pros" have of Poser... but that's a discussion I could rant on about for hours so I'm shuttin' up about it now...) scripting such as Mel and Maxscript are used quite extensively to improvise elements into the rigs that aren't necessarily available through the package outright. One of the reasons I love Maya so much is that literally everything in it can be codified and implemented through a Mel command line which offers a serious level of power over one's work. Unfortunately, I'm no Python guru, but I've found that there is actually a lot more "under the hood" of Poser before even hitting Python than usually gets talked about. Take ERC for instance: in our work, ERC in Poser is a direct reflection of how we use Set Driven Key scripting in Maya, even to such an extent that most of the time the control ratios are identicle! Actually, expanded ERC implementation is at the top of my wishlist for Poser 6: I would so LOVE to be able to ERC shader nodes and global attributes for things such as hair, lights, cloth settings... ERC-ability on that kind of stuff would open up some purely astonishing possibilities. Ah well; I'm degenerating into Poser daydreaming at this point, but suffice it to say that there is a LOT of "high end" paradigms that actually apply in Poser far beyond what the package usually gets credit for. -Les

I would so LOVE to be able to ERC shader nodes. You can. I don't have much time now, but I can share the details soon. -Stefan

That is something I'll be very interested in, considering our ongoing development of shape shifting humans (werewolves, werecats, etc). So far, we've had a lot of successful results animating the shifts with the full body morphs, and animated the transition from human to lycan textures using animated blend nodes, but if the blend node could actually be linked to the full body morph, well.... that would save a lot of work... would definitely have to update a few products with that technique worked out. ;) -Les

It's true that in a lot of "high end" environments (I put that in quotes, always, because IMO it's a term of snobbery that has done nothing but hurt the perception the "pros" have of Poser...

I'd like to echo that sentiment. But please know that when I say "high end" I'm simply trying to calssify, not to diminish Poser. It's a convention. I'm treated rather poorly by my classmates because I like and support Poser 5.

What most of the high end users don't realize, though, is how far this software has come, and how smart and innovative its users are.

Curious Labs is only a small part of the reason Poser has been so successful. It seems that the biggest part has been that very dedicated users have taken the time to come up with new technology and new ways of using the technology to push it forward.

Sorry I took you so seriously, btw. It may have just been me. Long nights. ;)

No offense taken. The Poser-biz can be full of long nights and misunderstood comments; comes with the territory. :) -Les

ERC materials: As soon as you click the little key next to a material parameter, Poser creates a separate parameter channel on the actor, visible in the graph editor and the parm dial palette. The good news is, this info is saved in a PZ2 as a regular parameter! All you need to do now is find your material parameter in the pz2 file and do the usual ERC editing. Email me if you want a sample file or need other help.