This might be an unforeseen use for Ragdoll. I don’t think @marcus anticipated this use for Ragdoll, but this is a WIP shot for my demo reel, and I’ve actually used Ragdoll to enhance the facial animations and lip-sync. This is all hand-keyed, using Metahuman faces in Unreal Engine 4, facial animations are all hand-keyed in Maya after which I ran a “Ragdoll Pass”.
I’ve also used it on the bodies for much of the secondary motion and “shakeyness” of the characters, anyone that has seen the film this audio is from will know hoe much Tom Cruise’s body shakes when he’s punching the air, I tried to replicate the same motion.
Thought I’d share this idea, as well as seeing if anyone can give me some feedback on how I can improve upon this, I’m kind of drawing a blank here when it comes to pushing this further, I don’t want to over-do it and over-think it and wind up with something worse than what I started with lol
This is pretty cool! was thinking about doing something like this with a lowres face mesh with points positions driving the face rig , I’d have distance constraints holding the points together.
Do you have any breakdowns showing the markers?
No I didn’t think of showing breakdowns when working on this, so I didn’t keep track of the progress, now it’s all on one merged layer and I can’t really go back. However keep in mind that Metahumans come with a control board when you bring them in Maya, the control boards is made of a series of cubes, and the way you actually export a finished animation is that you export the animation you have on that geometry and then (inside Unreal Engine) you import it to the same control board within ControlRig, (whereas normally you’d just export joint animation, but it doesn’t work for Metahuman faces)
So basically I turned all those cubes in the control board into dynamic objects set to worldspace, (so they wouldn’t fall to the ground) and just tweaking dampness and stiffness to get some nice secondary motion and jitters in the cheeks etc. it’s subtle, and it’s kind of hard if not impossible to interactively change stuff, basically what I have to do is look at how much the controllers in the sim are jiggling etc. and then record the sim, and look at the result in the face. It CAN get quite wild, especially with low dampness and low stiffness where the brows suddenly turn into mushy jelly, but I think it should be possible to apply ragdoll in such a way as to add the minute micromovements people make without even thinking about it, one step closer to realism and further away form the uncanny valley
That does sound pretty neat, Got any screenshots from an old scene? but yes you’re right anything that has transforms on could get some nice juicy dynamics applied
I mean I have the previous version of this shot (before I applied any Ragdoll. Actually before I even knew Ragdoll existed)
But the shot is kinda really bad at that stage and I feel shy sharing it lol