Automated Construction of Parameterized Motions

1
Automated Construction of Parameterized Motions

• Lucas Kovar
• Michael Gleicher
• University of Wisconsin-Madison

2
Parameterized Motion
Blend (interpolate) captured motions to make new
ones
Map blend weights to motion features for
intuitive control
(Wiley and Hahn 97 Rose et al. 98,01 Park et
al.02)
3
Adapting Parameterized Motion to Large Data Sets

• Previous work used manual blending methods on
  small, contrived data sets.

• We introduce automated tools that simplify
  working with larger, more general data sets
• Automatically locate examples
• Automatic blending (discussed previously)
• Accurate, efficient, and stable parameterization

Inputs one example feature of interest
4
Outline

1. Finding example motions
2. Parameterizing blends
3. Results

5
Outline

1. Finding example motions
2. Parameterizing blends
3. Results

6
Finding Motions

• Example motions are buried in longer motions.

ready stance
punch
dodge
punch
Strategy search for motion segments similar to a
query.
7
Related Work Searching Time Series Databases
Goal find data segments (matches) whose
distance to query is lt e.

• (Faloutsos et al. 94) place low-dimensional
  approximation in spatial hierarchy
• (Cardle et al. 03, Liu et al. 03 Keogh et al.
  04) motion data

Confuses unrelated motions with distinct variants
8
Logically Similar ? Numerically Similar!
9
Our Search Strategy
Find close matches and use as new queries.
Precompute potential matches to gain efficiency.
10
Determining Numerical Similarity
Factor out timing with a time alignment (just as
with registration curves).
Time alignment
Segment 1
,
Segment 1
Segment 2
Segment 2
Compare average distance between corresponding
frames with threshold.
11
Precomputing Matches Intuitions
Any subset of an optimal path is optimal.
Motion 1
Motion 2
Optimal paths are redundant under endpoint
perturbation.
12
Match Webs
Compute a grid of frame distances and find long,
locally optimal paths.
Motion 1
Represents all possibly similar segments.
Motion 2
13
Searching With Match Webs
At run time, intersect queries with the match web
to find matches.
Motion 1
Motion 2
14
Search Results

• 37,000 frame data set with 10 kinds of motion.
• 50 min. to create match web, 21MB on disk
• All searches (up to 97 queries) in 0.5s
• Manual verification of accuracy
• Can not discern meaning of motions!

picking up
putting back
15
Outline

1. Finding example motions
2. Parameterizing blends
3. Results

16
Natural Parameterizations
Blend weights offer a poor parameterization.

We need more natural parameters.
parameters
motion
reaching
hand position at apex
turning
change in hip orientation
jumping
max height of center of mass
17
From Parameters to Blend Weights
It is easy to map blend weights to parameters.

blend weights
blend
parameters
But we want !
This has no closed-form representation.

18
Building Parameterizations
Can approximate from samples
with scattered data interpolation (Rose et al
98).
Accuracy create blends to generate new samples.
(see also Rose et al 01)
19
Sampling
Require sampled weights to be nearly convex
and
for
Sample blend weights only for subsets of nearby
motions.
20
Scattered Data Interpolation

• Previous work uses an RBF interpolation
  method that does not constrain blend weights.
• (Rose et al 98,01) (Park et al. 02)

K-nearest-neighbor interpolation is (almost) and
ensures blend weights are nearly convex.
21
Outline

1. Finding example motions
2. Parameterizing blends
3. Results

22
Results
23
Discussion

• Parameterized motions make it easy to synthesize
  and edit motion.
• We want lots of them, so we need tools that
  simplify their construction
• Automated extraction of examples
• Efficient and accurate parameterization that
  respects boundaries implied by data