Noticed in the official SIGGRAPH 2022 Technical Papers Preview…
A 3D figure’s bone-movements drive momentary natural ‘swing-flicks’ in the dynamic clothing, e.g. a skirt flicking outward while dancing.
Have an AI take a single animation sequence, and automatically create new similar sequences that fit the same rig.
Add very realistic ‘white froth’ running on top of a stream of water, as an animation.
Take a product designer’s 2D concept sketch and automatically build and extrapolate a fully rotating 3D mesh for it.
Also a new AI to… “train primitives in seconds and render them in milliseconds, allowing their use in the inner loops of graphics algorithms”. Not 3D primitives, but Z-depths, “light fields, textures”. Sounds like a new type of ‘on-the-fly’ intelligent adaptive shader?
And a tool to “sketch objects at different levels of abstraction”. It’s “semantically aware”. So it sounds like an AI art-gen where you type “a cat sat on a mat”, and you get a picture of a cat on a mat. My guess is that here you’d get eight sketchy pictures of a cat on a mat, each more abstract than the next?
Finally, an interesting tool in which a five-year old child appears to ‘battle’ with an AI art-gen tool which has a robot drawing arm. They go through “a series of translational stages between humans and non-humans” in drawing, while presumably learning from each other along the way.