Enterprise AI Analysis
Designing Movement Generation Models in Collaboration With Voguing And Dancehall Dancers
Recent advances in Artificial Intelligence have enabled powerful generative models, yet few are tailored to dancers' practices. We present a long-term collaboration with a Voguing and Dancehall collective to design movement generation models trained on their repertoire. Our initial study with the dancers revealed that, despite limited physical realism, the generated movements inspired them. Iterative development led to Korai, an interactive tool for monitor- ing training, visualizing motion data, and prompting generation, which improved output quality. A subsequent structured observa- tion study compared three model variants with high, medium, and low fidelity to the original dataset's style. Results show that dancers favored either highly faithful or highly unfaithful outputs, rejecting medium fidelity as neither authentic to their style nor creatively stimulating. Our findings highlight how direct collaboration with dancers not only informs model design but also deepens under- standing of Al's role in supporting creative movement practices.
Léo Chédin, Laboratoire Interdisciplinaire des Sciences du Numérique, Université Paris-Saclay, CNRS, Orsay, France (leo.chedin@universite-paris-saclay.fr)
Jules Françoise, Université Paris-Saclay, CNRS, Laboratoire Interdisciplinaire des Sciences du Numérique, Orsay, France (jules.francoise@lisn.fr)
Baptiste Caramiaux, Sorbonne Université, CNRS, Institut des Systèmes Intelligents et de Robotique, ISIR, Paris, France (baptiste.caramiaux@sorbonne-universite.fr)
Sarah Fdili Alaoui, Creative Computing Institute, University of the Arts London, London, United Kingdom (s.fdilialaoui@arts.ac.uk)
Executive Impact: AI in Dance Innovation
Our collaborative research at the intersection of AI and dance has yielded significant advancements, translating into tangible benefits for creative practices.
0
Years of Collaboration
0
Dance Styles Covered
0
Interactive AI Tool Developed
0
Model Fidelity Levels Assessed
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Introduction
Methodology
Key Findings
Discussion
Conclusion
Introduction
The introduction highlights the historical context of computers in dance choreography, from early predictions by Michael Noll to recent advances in AI for movement generation. It emphasizes the gap in current Human-Computer Interaction (HCI) research regarding the integration of dancers' knowledge into AI model design and evaluation. The paper proposes addressing this by focusing on specific, codified dance styles like Voguing and Dancehall, in collaboration with practitioners.
Methodology
This section outlines a three-stage iterative design process in collaboration with Voguing and Dancehall dancers. Initially, a generative model (Chor-rnn) was trained on custom motion capture data. Feedback from improvisation sessions led to the development of 'Korai', an interactive tool for training monitoring and visualization. A subsequent structured observation study compared three model fidelity levels (high, medium, low) to understand dancers' perceptions.
Key Findings
Dancers found movements generated by the low-fidelity model (physically unrealistic but novel) creatively stimulating. High-fidelity movements (physically realistic and stylistically faithful) also inspired them, allowing for fluid improvisation. However, medium-fidelity movements (physically realistic but lacking stylistic legibility) were rejected as neither authentic nor creatively stimulating. This suggests an 'Uncanny Valley' effect in dance generation, where medium fidelity is less appealing than both low and high fidelity.
Discussion
The discussion introduces the concept of an 'Uncanny Valley of AI-generated movement' for dance, where physical realism alone isn't sufficient; stylistic legibility is crucial. The paper argues for greater dancer participation in AI system design, moving beyond passive observation to active co-creation, particularly for codified dance styles. It also highlights Korai's role in bridging the gap between AI development and dance practice.
Conclusion
The collaboration with Voguing and Dancehall dancers yielded insights into movement generation models. The interactive tool Korai facilitated the process. Dancers preferred either highly faithful or highly unfaithful outputs, indicating an 'uncanny valley' for medium fidelity. The study underscores the importance of direct collaboration with practitioners to design AI tools that genuinely support creative movement practices.
Iterative Design Process with Dancers
Curate Repertoire & MoCap
→
Train Initial AI Model
→
Improvisation Session
→
Develop Korai Interface
→
Refine Model & Evaluate
→
Structured Observation
→
Deepen Understanding of AI's Role
| Model Fidelity | Physical Realism | Stylistic Legibility | Dancer Reaction |
|---|---|---|---|
| Low-Fidelity |
|
|
|
| Medium-Fidelity |
|
|
|
| High-Fidelity |
|
|
|
Uncanny Valley
Concept of AI-generated movement where medium fidelity is less appealing than low or high fidelity, due to lack of stylistic legibility.
Korai: Interactive Tool for AI-Dance Collaboration
Problem Addressed
Initial AI model development workflow lacked interactivity for training, generating, and visualizing movements, making qualitative assessment difficult for dancers and researchers. Manual rendering with Blender was tedious and non-real-time.
Solution Implemented
Korai, a web-based application, was designed to provide real-time 3D rendering of movements, interactive seeding for generation, and comparison of multiple model outputs. It enabled researchers to monitor training and allowed dancers to explore generated movements directly.
Impact
Korai significantly improved the feedback loop between AI developers and dancers, streamlining the iterative design process. It facilitated easier visual assessment of movement quality and allowed for dynamic exploration of generative models, fostering deeper insights into dancer-AI interaction.
Advanced ROI Calculator for AI Integration
Estimate the potential return on investment for integrating AI-powered creative tools into your enterprise operations.
Estimated Annual Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
A phased approach ensures seamless integration and maximum impact of AI in your creative workflows.
Phase 1: Data Curation & Model Training
Collaborate with dancers to curate specific movement repertoires (Voguing, Dancehall), perform motion capture, and train initial AI generative models using this bespoke dataset. Focus on achieving basic movement continuation.
Phase 2: Interactive Tool Development (Korai)
Design and implement 'Korai', a web-based interactive interface for real-time visualization of 3D movements, interactive seeding, and parameter control. Enable comparison of different generated sequences.
Phase 3: Iterative Model Refinement & User Studies
Conduct structured observation studies with dancers, comparing models with varying fidelity levels (low, medium, high). Gather qualitative feedback on stylistic legibility, physical realism, and creative inspiration to refine model parameters and architecture.
Phase 4: Integration into Choreographic Practice
Explore integrating refined AI models and Korai into long-term choreographic creation and improvisation residencies. Investigate how dancers use the tools over extended periods and in diverse artistic contexts.
Ready to Transform Your Creative Practice?
Partner with OwnYourAI to design and implement bespoke AI solutions tailored to your unique artistic and enterprise needs.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat