I am a Tree: Embodiment Using Physically Based Animation

20000. 25000. 30000. Stressed. Sad. Neutral. Happy. Cool. Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives ...
2MB taille 8 téléchargements 262 vues
I am a Tree: Embodiment Using Physically Based Animation Driven by Expressive Descriptors of Motion Caroline Larboulette, Sylvie Gibet EXPRESSION team, IRISA lab University of South Brittany MOCO 2016, Thessaloniki, Greece

Motivation How do motion qualities influence the evolution of a 3D simulated system ?

Captured Human Movement

Non-Anthropomorphic Entity 2

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Related Work • Control of a MSS using motion descriptors [Alaoui et al. 13]

• Dance with virtual body extensions [Bisig et al. 16] • Sonification of human full-body movement qualities [Kolykhalova et al. 16]

3

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Our System • Physically based system • Mass-Spring System: 5 particles + 9 springs • Leaves: 1 particle emitter for each branch • Rendering: • Trunk: static generalized cylinder • Branches: Hermite Splines + cylinders 4

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Motion Descriptors • • • •

CoM: distance from Center of Mass to Root joint Bounding Volume: volume of the bounding box Shape Directional: arc-like versus spoke-like Extensiveness: normalized sum of the distances between CoM and end effectors (hands, feet, shoulders, head)

5

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Motion Descriptors • Laban Effort Descriptors • Flow Effort: free versus bound • Time Effort: sudden versus sustained • Space Effort: direct versus indirect • Weight Effort: strong versus light • Equations can be found in A Review of Computable Expressive Descriptors of Human Motion, Larboulette and Gibet, MOCO 2015 6

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Tree Shape • Bending of the tree - CoM-Root Pcenter Mass-Spring System (3D)

7

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Tree Shape • Bending of the tree (CoM-Root Pcenter MSS) • Breathing of the tree - Volume of Bounding Box Spring restlengths

8

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Tree Shape • Bending of the tree (CoM-Root Pcenter MSS) • Breathing of the tree (BVol Spring L0) • Curvature (arc-like versus spoke-like) - Shape Directional m1.z values of Hermite Splines

9

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Tree Shape • Bending of the tree (CoM-Root Pcenter MSS) • Breathing of the tree (BVol Spring L0) • Curvature (Shape Directional m1.z Splines) • Fluidity (free versus bound) - Flow Effort Spring Stiffness and Damping 30000

Stressed

25000

Sad

20000

Neutral

15000

Happy

10000

Cool

5000 0

0

1000

2000

3000

10

4000

5000

6000

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Leaves • Time Effort (sudden versus sustained) vz : how fast leaves fall rate of particles emission: how fast leaves rise • Space Effort (direct versus indirect) vx & vy : How straight leaves fall

11

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Parameters Mapping • Colors • Extensiveness Hue • Weight Effort Brightness strong = dark ; light = bright

12

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Results

• All together ! • Placed on the arousal / valence graph, according to the expected values

13

Arousal

Stressed

Sad

Happy

Aroused

Valence

Calm

Unpleasant/Pleasant

Cool

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluations • Baseline Experiment • 5 animations, 17 participants • Asked to qualify each video • Emotions were recognized • Valence / Arousal Perceptual Study • 30 videos: 3 movements x 5 emotions x 2 renderings • 31 participants • Valence and Arousal graded on a 1-7 Lickert scale

15

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluation: Happy

Expected Scores: Positive (valence > 4) Energetic (arousal > 4)

Valence Arousal 16

Character

Tree

All

5,366 5,258

4,258 5,731

4,812 5,495

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluation: Cool

Expected Scores: Positive (valence > 4) Lethargic (arousal < 4)

Valence Arousal 17

Character

Tree

All

4,28 3,796

3,978 4,28

4,129 4,038

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluation: Neutral

Expected Scores: Valence = 4 Arousal = 4

Valence Arousal 18

Character

Tree

All

4,71 3,645

3,86 3,763

4,285 3,704

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluation: Sad

Expected Scores: Negative (valence < 4) Lethargic (arousal < 4)

Valence Arousal 19

Character

Tree

All

3,602 2,312

3,29 2,946

3,446 2,629

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Evaluation: Stressed

Expected Scores: Negative (valence < 4) Energetic (arousal > 4)

Valence Arousal 20

Character

Tree

All

4,559 4,914

3,559 4,602

4,059 4,758

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Conclusions • Successfully coupled expressive motion to a simulated 3D system • Indirect mapping with higher level descriptors • Non-anthropomorphic physical system • Perceptual studies showed that emotions were transferred

21

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Conclusions • Need to investigate different systems • New physical model • New set of descriptors • New mapping • Turn our experiments into live performances

22

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Perspectives 1. Descriptors Sensitivity • Parts of the body used for computation • Time window and offset • Nature of the skeleton’s structure and size 2. Mapping Importance • Effect of each parameter • Normalisation of the descriptors wrt the simulation 23

Related Work - System/Descriptors - Mapping - Results/Evaluations - Conclusion/Perspectives

Perspectives 3. Structure of the Physical System • Anthropomorphic / Non-Anthropomorphic • Methodology: do partial evaluations • Watch the variation of a descriptor in real-time • Do mappings for unitary simulation (one object, one parameter at a time)

24

Thank You ! ευχαριστώ !

Caroline Larboulette, Sylvie Gibet University of South Brittany / IRISA http://www-expression.irisa.fr/

25