So, for somebody more technically-minded, what's the merger process between the mocap here and the
Ziva Dynamics AI-assisted animation systems mentioned earlier regarding this character? Am I reading right that the mocap done for the keyframes, and then Ziva FX was used for the skin and musculature?
https://www.neogaf.com/threads/senu...-create-that-40ft-film-quality-troll.1625831/
https://mp1st.com/news/senuas-saga-...rtists-to-create-that-40ft-film-quality-troll
https://zivadynamics.com/blog/senuas-saga-hellblade-ii-gameplay-reveals-character-by-ziva-dynamics
When I first read that whole "1 Month for 40-foot Film-Quality Troll", I thought what they were doing was more an Physics-Based Animation / IK solution, where they built and trained AI to imagine how a real 40-foot-tall being would move and struggle under its own weight and bulk. As in, once you have a big creature like that and have it struggle like a baby for "years" in animation training, eventually you'd have a simulated creature that would know how to roll and stretch its body around to position itself or how much arm strength and pushing of legs it'd take to move its massive body; how a tremendous creature like this with the mass of several elephants would move itself in as natural a way as possible. Once you have a creature that has the "muscle memory" to move in the way that it would know how, you would then add actions you want it to do (crawl through a cave, smash through a barrier) as well as facial animations, and maybe you'd have to tweak things further if the AI's understanding of the action doesn't meet your artistic needs, but otherwise, let the computer do the animation work, because nobody has actually seen a 40-foot giant before, and a computer might know better how it'd physically move than a person could guess.
...But then, they just mocapped the whole scene anyway, and just did the usual thing of slowing down the motion ala old Godzilla films for a "Slow = Giant" effect.
So, for my level of excitement of the technical achievement here, there's some pluses and minuses for me. On one hand, I thought they were doing the giant "for real", not using traditional mocap and keyframed acting/animation to simulate the beast... it's just a guy on the floor, pretending he doesn't have a leg, trying to move like he's made of blubber. That seems pretty normal to me, they've been doing mocap acting for a long time, such as this
similar mocap session with a big creature. (Plus, I feel like they really should have built him a "fat suit" or like filled the suit with pennies or something, so that the struggle was real, instead of him puppeteering himself to act like he was heavy.) On the other hand, the fat and muscle and skin simulation is pretty extraordinary on the troll, and although 1 month of 3 staffers' work on one creature seems like a lot (plus all the AI time training itself on understanding the meat physics,) hopefully the time spent on this made further creature designs and enemy encounters move more smoothly once they got through this challenge.