AI video generation tool that creates videos from text prompts, images, or reference clips with strong character consistency.
Edge Dance

About Edge Dance
Edge Dance is an AI tool that creates high-quality choreographies from music. It uses a frozen Jukebox model to encode input music into embeddings, and a conditional diffusion model to map the embeddings to 5-second dance clips. Temporal constraints are applied to batches of clips to ensure temporal consistency before stitching them into an arbitrary-length full video. Edge Dance is the perfect tool for anyone interested in creating stunning dance videos from their favorite songs. The AI-generated choreographies are smooth and expressive, and the interface makes it easy to craft a unique dance that suits any style or genre. With Edge Dance, dancers and choreographers can quickly put together professional-level videos with minimal effort. From hip-hop to jazz, and from contemporary to Latin, the possibilities are endless. Whether you’re a seasoned dancer or a novice, Edge Dance has something for everyone.
Key features
- Generate high-quality choreographies from music
- Create stunning dance videos effortlessly
- Craft unique dances in any style or genre
- Smooth and expressive AI-generated choreographies
- Easy-to-use interface for crafting unique dances
- Professional-level video creation with minimal effort
Use cases
- Creating dance videos for music performances
- Developing professional-level dance content for films, TV shows, or commercials
- Crafting unique and engaging dance routines for events or exhibitions
Pros
- Generates realistic and physically plausible dances from arbitrary input music
- Supports editable synthesis with joint-wise and temporal constraints for customization
- Maintains temporal consistency across batches to produce arbitrary-length dance videos
- Trained with physical realism in mind, reducing unintentional foot sliding
- Human raters strongly prefer dances generated by EDGE over previous methods
Cons
- Limited to 5-second dance clips during training, requiring temporal constraints for longer sequences
- Dependent on Jukebox for music embeddings, which may restrict compatibility with all music types
- Complexity of editable constraints may pose a learning curve for novice users
Frequently asked questions about Edge Dance
What does Edge Dance do?
Edge Dance generates realistic, physically-plausible dance choreographies from input music using a transformer-based diffusion model and Jukebox for music feature extraction.
Who is Edge Dance suitable for?
It is designed for dancers, choreographers, and creators looking to produce high-quality dance videos with minimal effort, supporting various styles and genres.
How does Edge Dance ensure temporal consistency in generated dances?
It applies temporal constraints to batches of 5-second dance clips, enforcing continuity between sequences to create arbitrarily long, coherent dance videos.
Can Edge Dance generate dances of any length?
Yes, by imposing temporal constraints on batches of sequences, Edge Dance can produce dances of arbitrary length while maintaining physical plausibility.
Does Edge Dance support editing or customizing generated dances?
Yes, it offers joint-wise conditioning, motion in-betweening, and dance continuation, allowing users to edit and refine dances to their preferences.
How does Edge Dance handle physical realism in dances?
It uses a Contact Consistency Loss to learn when foot sliding is intentional, improving physical realism while preserving natural dance movements.