Rigging

RIGGING

REALISTIC HORSE RIG

Hello, I am proud to show you one of the project I got the occasion to do this year. It was a great deal of getting the anatomy of a horse right (did you know they could have varying numbers of vertebraes?), and to make some intuitive controls for the animators to move it freely. 

One of the challenges of this Rig was to actually make some constraints and some automations so that the movements would always be as close to real life as possible, because for the one of you who know a bit more about Rigging, this is just an animation rig meant to then be used for a muscle rig, a Ziva system which is still in preparation for me. 

SCRIPTING

But this specific project was not only great to learn more about the animal's anatomy, it was also a good excuse to get to dive a bit more into scripting. Indeed some of its features were added via script made on the spot, to either automatically create a coherent set of joints for the ribs (because yes for this type of thing you need at least one joint per bone).  Also some more day to day tools like changing the color or shape of a controller faster or ceate some attributes automatically for all the "ankle" controllers. 

THE FINAL TEST

And then came the final test: giving the rig to the animators to test it out. And if all of them liked it and found it efficient, two problems came around : 

First a rig needs to be intuitive, if all the controllers are the same size and same color we can't expect the animator to know what to use and how from the get go, because, even in rigging, style has its importance. 

And second, some issues occured in the animators scenes that did not occure in my scene, and that was due to a simple problem, the more the animators manipulate the rig, the more likely it is that they will move something that is not supposed to be touched. And that's absolutly logical, one of the other goal of a rigger is to make the rig unbreakable, the best way to keep anything that shouldn't be touch from being, well, touched, is to make it unreachable. So after the rig is done it is time to lock everything and hide all the attribubutes that should not be used. 

And then, finally, you have something practical, and stable to give to your team. 

FACIAL RIG

Here is another project I got to work on with a wonderfull team on the short movie "Don't be Tuna'ive", I rigged the whole character but let's focus on Takashi's face.

Facial Rigging, which in my opinion, for a regular biped, should take at least as much time as the rest of the body. The face is the best vector of emotion we could ever find, which is great but we also see it everyday at from so close. We're so used to seeing faces that we can recognize them in places where there's none. Which is great in the day to day life but very challenging for a rigger. As normally, a LOT of the facial muscles are interconnected, for exemple when you smile your eye muscles get involved and this has to be transcripted through the rig of the character. 

TECHNICAL DIFFICULTIES

Since a lot of very precise movements are involved a same vertices will be influenced by more controllers than the vertices on the rest of the body, which is fun, knowing that the face usually has the highest density of the body. To ease out this tedious task I used ngSkintool in a very specific way: I did a first path in black and white putting white values for the most influencing joint of the zone. 

Then it's all a work of transition, and my technique to not get a "too smoothed out" effect, was to get the add brush set at a low values and enlarge the zone of influence of the joint to what I want it to be at the end, then i'd do that to all the other joints until none of them has a full influence anymore. Be carefull though, NEVER remove wheights during this process, and honestly try not to remove anything as much as possible. Once this is done, I grab my smooth brush and tick "distribute to other influences".  This will avoid having weights being distributed automatically to unwanted joints. 

But even though a good skinning will solve most of your issues, it is never be perfect by itself, that's why you can use a deltaMush deformer to smooth out some specific parts of the body. 

And finally, to add some more dynamism to the face, one crucill step is blendshaping, here, for this project, I added some corrective blendshapes activated on only some values for specific controllers, to get more volume preservation. And to go further I created some combination shape to keep it all coherent.