View Full Version : Ultimate Dynamic Ragdoll Rig
07 July 2005, 02:32 AM
After some time in secret development (~4 months) I have devised a fully articulated dynamic rig for maya that supports collisions and is fully robust. I have seen some talk regarding these types of skeletons and I've seen simulations using rigid bodies. Sure the simulations with rigid bodies have their place, but this is something completely different.
It's about 15 megs.. quicktime with sorenson compression..
Using a control skeleton we can drive a dynamic skeleton and have it react dynamically to all our motions. Please have a look at the quicktime movie. Any feedback is appreciated as well as comments. I'm not sure where to begin in explaining it, but I'd love to hear any questions or comments.
<shameless_plug> As well I must say I'm totally accepting job offers in the TORONTO area.. I have a bachelor's degree (Computer Science) from U of T and I've just completed a diploma in computer animation (Maya/Shake) from the Toronto Film school. I'm putting together a demo reel of my best stuff but basically I can rig just about anything and I'm quite well-versed in MEL script for UI's and simplifiying complex repeatable actions.. plus I love dynamics. </shameless_plug>
By the way.. here's a render of the actual character.. and a screencap
07 July 2005, 10:52 AM
i liked it, impressive stuff.
07 July 2005, 01:09 PM
Thank you.. The truth is I'm not sure exactly how useful it is, but it brings together a lot of concepts I've seen floating around sort of like on their own. The trick is how to transfer the shape of the control skeleton to the dynamic skeleton without losing any degrees of freedom and as well without causing double transformations. If you set out to make a similar thing using the spline approach from the documentation or 'Maya secrets of the pros' you'll quickly discover that for every new limb you try to add (like arms as 'children' of a spine, or fingers as children of the arms) you'll be getting a double transformation. That had me buggered for about a month and I almost gave up, but as you see there is a way to resolve this issue.
Not to mention that with a spline you will usually risk having the limbs pop off under acute angle transformations, so although it may seem to work around the origin and in rest pose, in many cases it will result in the limbs popping off the splines at odd angles.
07 July 2005, 01:50 PM
Hey Alex, the rig is looking wonderful. I have a couple of questions of you:
* How much the dynamic effect can be hand controlled (attribute like stiffness, etc)?
* Is there a way to pin some parts so that the will not be affected?
07 July 2005, 02:49 PM
It's great to see people like you making researches in order to improve the CG industry! Best wishes man :thumbsup:
07 July 2005, 03:18 PM
That is absolutly amamzing. Great Job!
07 July 2005, 04:55 PM
That had me buggered for about a month and I almost gave up, but as you see there is a way to resolve this issue.
That shows one very important trait, you stuck with it, I do the same, I'll be thinking of a problem in the back of my head while I go about other tasks, and if you stick with it and don't give up, you'll succed.
I think m8, its fair to say that you stand a pretty good chance of being hired from the development of this rig.
07 July 2005, 06:09 PM
The rig is really great! I am curious to see how the animation of the rig would turn out so an animation test would be good or even create a short based on it. A feedback from any seasoned animator after using it could maybe shed some light on its practical use for doing animations.
07 July 2005, 11:42 PM
It seems that this rig is light years ahead of anything else out there at the moment. I certainly think this is a step in the right direction and could positively change the way we approach animation in the future. I would guard this discovery with my life!
07 July 2005, 01:15 AM
Wow, that was really impressive! That rig must be soo much fun to play with! :)
07 July 2005, 02:45 AM
:applause: u know wat the way u made was great u are a exparty maybe u can teach me how to do it :)
07 July 2005, 07:29 AM
That looks really cool, but I'm not sure how useful it is, since Maya will end up putting a lot of secondary animation into your scenes. So all your animations will probably look wobbly and funny, after all.
Great work, though, shows again that you just need a screwdriver and some bubble gums to develop a spaceship ;)
07 July 2005, 09:44 AM
I second the call for some real animation tests with it...characters with emotions..because that way we could really see something cool at work. Some wonderful principles at play here.
I certainly think this is a step in the right direction and could positively change the way we approach animation in the future. I would guard this discovery with my life!
Actually, there's a great saying: If you want someone to pay attention to your great new discovery, it won't work to keep it hidden...you've got to shout it from the rooftops to be heard.
07 July 2005, 08:31 PM
Overall I'd say the concept and idea is a good one. The thing that sticks out in my mind is its actual use in a production environment. You don't really see alot of RagDoll effects being done in production as most animators prefer to handkey this type of thing to give a more 'dramatic' feel. I'd be interested in seeing some actual animation tests like a walk cycle, run cycle or a jump to see how it works under those conditions :)
On another note, do you have any other work? As a TD/Rigger myself, I would suggest showing off more organic stuff that involves weighting if you are actively looking for a job around this field. Anyone can place joints, all comes down to your weighting skills and the ability to really take the character to another level :thumbsup:
07 July 2005, 05:44 AM
It does seem that in production we stick to first principles and I totally agree with this approach. we don't need like to send a hair rig down the pipeline and have it frizz up at render time because of any million reasons like missing cache files or simple inexperience with hair rigs which is completely understandable, not to mention using hair for a purpose other than which it was intended.
So as I sit at home and munch on smarties, I'd like to elaborate on the rig we have before us and for whose analysis I am very grateful to all of our colleagues. Your concerns are well founded, and I propose that it is indeed an oddity and it won't ever replace frame-by frame animation which is the only real way to do things anyway (or maybe two-frame by two frame like stop motion, but I digress...)
This rig was designed to supersede the zoids rig in "maya secrets of the pros' and then I thought 'wouldn't it be nice if they were kinda rubbery.. because I find most animation lacks that extra level of super finesse.. like quaternary animation of you will. Sucky animators achieve primary animation, good animators have secondary animation, great animators have tertiary animation but four levels are rare indeed. Also I find that much of it has a rushed quality like bending the laws of physics to suit the impatient demands of the deadline. (I am gingerly side-stepping discussions of motion capture with my apologies to Dr. Aki Ross)
Anyway.. I find that over-all working with these rigs is like animating MUPPETS.. it's a different look than we're used to in CG.. it does smack a bit of 'look-what-I-can-do' rather than eye-candy but I'd totally fly to the moon on a bubblegum rocket as it was so eloquently put. I'm gonna keep this thread alive not with idle chatter but with solid animation tests so I'm retreating to the lab gents and I'll hope to have something reasonably fetching in the next few days or so!
As an aside: I am very grateful for all of this discussion and all of the opinions which I have totally taken to heart. Anyway there's no wasted research but like heinz may have a ketchup-flavoured ice-cream in production for all we know. or a mustard hemorrhoid cream. I hope this will be on-par with umm.. nutella sandwiches at least!
07 July 2005, 06:25 AM
How would this help when you key do you have to "catch" the secondary motion and key is quickly to animate the dynamic movement? Does it rect differently based on how it is keyed and the type of keys/splines used when animating?
07 July 2005, 06:50 AM
it unclear if this "ragdoll" can animate with collisions or forces. Check out Endorphin
for some additional ideas. good luck
07 July 2005, 03:36 PM
I love this idea, itīs pretty creative.
Your tool can rig any existing rig?, I mean a "custom one"?
Thankīs, and please, ask someone to animate a simple walkcycle for this guy! :D
07 July 2005, 11:34 PM
07 July 2005, 12:28 AM
interesting concept. i can see the cool factor in this, but i'm not sure if i can see where it would be useful in a production. maybe if a character was underwater? :)
the secondary would definitely have to be a keyable, controllable element in any case. perhaps a slider to say 'ok, blend this from 0 to 1 where 1 is snapping back to no more jiggle/lag'.
i'd also say that its not really a ragdoll. ragdolls you can grab from any joint and toss around, and the rest of the joints will react dynamically. for this to be ragdoll, you'd have to have something like a fullbody IK (think motionbuilder humanIK) going on... gravity basically would have to take over on all controls other than the one that you have selected.
i would certainly like to see some animation with this though. it could totally turn out to be a crazy, eye catching, style of animation. think of frame blending/cartoony stretching like chicken little...
anyway, great development!
07 July 2005, 09:45 PM
Interesting work. While I don't know if this type of behavior would be useful for most character rigs, I can definetly see it's use in animating natural phenomena (like a tree in a hurricane).
07 July 2005, 08:43 AM
Sorry for the delay and thank you very much for all the input. I totally agree that this is not technically a 'ragdoll' setup and I don't think it's completely feasible to implement a motionbuilder-style 'human-IK' with the current state of maya, as that would violate the hierarchical rules of the skeletal structure though I'm sure it could be hacked together one way or another, one is certainly better off using motion builder. Now of course where does that leave maya as an animation tool? that is certainly a good topic for another thread.
As for control, there is a very fine level of control of all the dynamic features because the rig is based on hair. I would like to offer the following few small animation tests, all based on the same animation curve data, but with some modifiers of the outcome due to dynamics.
Please click on the images to view the animations.
example A: THE ORIGINAL ANIMATION
example B: THe original, with dynamics turned ON
example C: An AIR field added to push him from behind, since the rig responds to dynamics.
example D: a bumpy ground added that is set to collide with the legs. Now the legs aren't moving high enough to stay on top, so you see they drag as if he's stuck in sand or something,.
You may also be interested in seeing him swinging from a tree:
Now I must reiterate that all the above animations are based on the same STATIC animation, and then there are added dynamics including fields and collisions. I don't suppose it's common to add fields to affect keyframed animation but as you see it's entirely possible.
I'm leaving these without comment for now, but I must say it's a look that's quite unusual in CG as somebody posted, and yes I do believe it would do for trees in a hurricane, or like devils in a sandstorm.
Thank you very much for your interest in this thread.. dreadfully sorry for the delay, it was a technical glitch unrelated to this project that kept it on ice for a few days longer than anticipated.
07 July 2005, 11:41 AM
the last demo is a bit morbid LOL ...but i think that ones outlines the power and potential of your rig ...which i might say is very inspiring and innovative ...congrats mate!
As per your cycle tests, its probably fair to say that this kind of rig is not really best suited to a more 'traditonal' setup ...however, for things like a hangman ;) and prop rigging in a dynamic environment (as mentioned for a hurricane etc) and for specific shots that call for this kind of fluid / overlapping motion , even if it is just for a portion of a character, is an option!!
So, in my HUMBLE opinion, great skills! Fantastic innovation, and has strong usefulness in the right areas.
07 July 2005, 07:07 PM
Brilliant work littlepixel. Those animations were quite interesting. Now perhaps you can shed some light on how you actually accomplished this type of behavior.
I know how to control joint chains with dynamic hair curves. The problem of affecting them from both ends can be solved by using a hair constraint. But how do you get a completely separate limb to affect a different part of the body? I guess what I'm asking is how do you interconnect the different sections?
The way I see it, if this process could be automated quite quickly (say with a MEL script), then this would definetly become a feasible manner with which to rig all sorts of things. I'm sure all of us here would appreciate some more details. Thanks for the cool stuff and keep up the great work! :)
07 July 2005, 07:07 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.