AI animation support? ("Sam walks up to Hank and slaps him on the back")

NVIDIA has some features like path animations (get a character to walk along a path). Any plans to have navigation support (walk around objects in room like tables) so you can use ChatGPT or similar to generate an animation clip from text such as “Sam walks up to Hank and slaps him on the back”? Obviously you have to tell it the starting point, but there is

  • plan the route through furniture and walls
  • work out the right place to stop
  • blend the animation from walking to stopped
  • slap character on back animation (blended from idle pose?)
  • does other character react (move) when slapped on back

E.g. with “digital humans” effort, how far is that likely to go? There were some early videos doing text to animation - is that work continuing?

What a great scenario. That sounds like a great extension that someone should make. If you’d like to make it, I think you’d find enthusiastic support in the Omniverse discord channel.

There was an old NVIDIA video about generating a single animation clip from multiple movements. Sounds like that is not an active area of development then?