1
|
- Michael Gleicher
- and the UW Graphics Group
- University of Wisconsin- Madison
- www.cs.wisc.edu/~gleicher
- www.cs.wisc.edu/graphics
|
2
|
- High Quality, Expressive Motion
- Need motion capture (examples)
- Flexible, long-running, controllable
- Synthesis from Examples!
|
3
|
- Flexibility:
- Link motions to make sequences
- Blend motions to gain control
- Use Databases of Examples:
- Find related motions in databases
- Combine data for interactive systems
|
4
|
- Motion Graphs
- Link motions to make long sequences
- Snap-Together Motion
- Synthesis for interactive systems
- Match Webs
- Find related motions in a database
- Registration Curves + Parametric Families
- Combine motions to make spaces
- Plus some others…
|
5
|
- New motions from pieces of old ones!
- Good news:
- Keeps the qualities of the original (with care)
- Can create long and novel “streams” (keep putting clips together)
- Challenges:
- How to connect clips?
- How to decide what clips to connect?
|
6
|
- Transitions between motions can be hard
- Simple method work sometimes
- Blends between aligned motions
- Cleanup footskate artifacts
- Just need to know when is “sometime”
|
7
|
- Factor out invariances and measure
|
8
|
- “Undo” the differences from invariances when assembling
- Rigidly transform motions to connect
|
9
|
- Find Matching States in Motions
|
10
|
|
11
|
|
12
|
- Find many matches (opportunistic)
- Good: Automatic
- Good: Lots of choices
|
13
|
- Any walk on the graph is a valid motion
- Generate walks to meet goals
- Random walks (screen savers)
- Search to meet constraints
- Other Motion Graph-like projects elsewhere
- Differ in details, and attention to detail
|
14
|
|
15
|
- Given a path
- Find a motion that minimizes distance
- Combinatorial optimization
|
16
|
- Search the graphs for motions
- Look ahead avoids getting stuck
- Cleanup motions as generated
- Plan “around” missing transitions
- Optimization gets close as possible
|
17
|
- Different set of tradeoffs!
- Runtime must be:
- Willing to sacrifice quality to get
|
18
|
- Search: Look ahead to get where you need to go
- React: Always lots of choices. Something close to need.
|
19
|
|
20
|
- What if motions matched exactly?
- Match both state and derivatives
- Match reasonably at a larger scale
|
21
|
- Add in displacement maps
- Bumps we add to motions
- Modify motions to common pose
- Compute changes at author time
|
22
|
- Pick set of match frames
- User selects
- System picks “best” one
- Modify motions to build hub node
- Check graph and transitions
|
23
|
|
24
|
|
25
|
|
26
|
- Graphs provide discrete choices
- Use pieces of the database
- Can’t capture ALL examples
- Synthesize new motions between example by blending
|
27
|
|
28
|
|
29
|
|
30
|
- Encode the relationships between similar motions
- (video of pair blending apps)
- If we have a big database…
- How do we find similar motions?
- How do we use several examples?
|
31
|
|
32
|
|
33
|
- Match Webs
- Search for similar motions
- Registration
- Align motions for blending
- Parameterization
- Sampling
- Improve nearest neighbor interpolation
|
34
|
- Example motions are buried in longer motions.
|
35
|
- Motions can be different lengths.
|
36
|
|
37
|
|
38
|
- Distance between corresponding frames (in the best time warp)
- Factors out timing differences
- Allows arbitrary distance metrics for frames
|
39
|
|
40
|
|
41
|
|
42
|
- 37,000 frame data set with ten different kinds of motions.
- 50 minutes to compute match web
- 21MB on disk
- All searches (up to 97 matches) in ≤ 0.5s
- Manual verification of accuracy
|
43
|
|
44
|
|
45
|
- Accuracy: create new blends to get additional
|
46
|
|
47
|
- To the UW graphics gang.
- Animation research at UW is sponsored by the National Science
Foundation, Microsoft, and the Wisconsin University and Industrial
Relations program.
- House of Moves, IBM, Alias/Wavefront, Discreet, Pixar and Intel have
given us stuff.
- House of Moves, Ohio State ACCAD, and Demian Gordon for data.
- And to all our friends in the business who have given us data and
inspiration.
|