Project discussion

Note: a name next to a topic doesn't mean anything other than the fact that I think I remember that person saying something about the topic; it doesn't mean you committed to anything. However, I would like to start associating names to topics. At this point you can be associated with more than one topic.

I think we would have a better chance of getting something done and publishable if we formed a small group for each project. That doesn't mean you can't work alone - but I'm not sure if something useful can be done by one person in what's left of one quarter for most of these topics.

• non-photo-realistic animation (Cheng): How can motion, force, be represented in the rendering? Motion blur, stretching in the direction of motion are simple examples of this (although motion blur is really 'realistic' as far as a camera model is concerned, it's still some extra processing required to represent motion).
  • are there other interesting techniques?
  • is there some 'theory' that can be developed about this?
  • is it useful? it might be for artistic purposes. Does it help in visualization?
  • there should be a paper possible. unknown whether there is anything beyond that.
• physics of skater (Cheng, Youri): modeling the dynamics of an articulated skater on ice
  • modeling in full 3D might get prohibitively complicated. The ice simplifies a lot of things, but the blade cutting into ice creates complications.
  • a 2D top-down simulation might be a viable demo
  • I'm not sure there's a paper in this.
• character modification (Oleg): given a keyframed walking character, how can the gait and secondary behavior be automatically modified to represent an [older, younger, heavier, tired, energenic] person?
  • literature: retargeting for non-mocap characters
  • a lot of work has been done in retargeting - this is a little different and can use results from retargeting
  • can a reasonable tool for animators be developed
  • need to bite off a chewable chunk
  • paper possible if a nice combination of attributes can be identified
• crowds (Neelima, Ying, Matt)
  • looking beyond pedestrian scene, transportation stations, and panic
  • what about crowds that interact - shove, push, trample
  • security camera coverage when crowds are present (matt)
  • crowd congestion during searching - also outdoor scenes (matt)
  • security for, say, presidential candidate
  • position guards - what's visible?
  • what interesting questions can crowd simulation answer?
  • paper possible if a nice application can be identified
• multi-touch table (Jae, Matt)
  • it would be nice to come up with a nice application for this - maybe educations
  • cooperative scrabble (matt)
  • some kind of gaming interface
  • paper possible
• modeling non-standard body types (Ben): most CG human characters are in shape and 30-ish. what about everyone else?
  • different motion based on age, sex, weight
  • Meg Geroch started on this and had a nice direction for a Ph.D., but ended up retiring before finishing.
  • paper possible as well as continuing research
• construction (Steve): simulating a complex process
  • e.g., simulate traffice of materials and tasks on-site
  • use spatial reasoning to determine bottlenecks, conflicts
  • may be too complex to get a handle on - can a simple scenario still be interesting to work on?
  • need graphics for what? visualization - what about geometric reasoning like for construction site traffic
  • paper possible
• fluids (Sundaresan,Matt):
  • Stam's code is available and fun to implement - it's 2D
  • lot's of work on fluids - might produce nice demo
  • there's been a lot done already on fluid flow (6 academy awards for technical achievement having to do with fluid flow this year)
  • need a nice twist to what has already been done; how about top down river simulation? maybe for erosion study - paper? need to think about this
  • flood simulation - interesting to blend fluid with other animation (matt)
  • paper possible if a nice application can be identified
• Foley (Oleg): automatically generate sounds to go with an animation
  • what kind of tool would animators want
  • game technology does this in real-time
  • what are the issues for off-line animation?
  • Ben pointed me to this paper. a quick scan of the paper tells me that it still needs to be transformed into a tool for animators.
  • produce sound track in Maya or produce tags for use by animators in post-production
  • paper if there are interesting issues separate from what is done in computer games
• modeling a particular person
  • Thalmann's did a movie of Bogart and Marilyn Monroe awhile back
  • Capturing this for cartoons is here (thanks for the pointer, Ben)