Feedback
All Tools
Browse AI tools
Free AI Video
100% Free AI Video Generator
Free GPTImage2
Best Image Generator
Seedance2.0
The Future of AI Video Is Here.
Free AI Image
Truly Free AI Image Generator
Gemini Omni
Gemini Omni Video Generator
Veo3.1
Create Stunning Videos with Veo3.1
Kling3
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
100% Free AI Video Generator
100% Free Image Generator
Seedance 2.0 AI Video Generator — The Future of AI Video Is Here
Gemini Omni Video is Here
Kling 3 Is Here
Kling 3 - See the Sound, Hear the Visual.
AI Video Effects
AI Effects - Create Funny Videos Easy!
Wan Dancer — Music-to-Dance AI Video Generator
Turn a single portrait photo and a music track into a beat-synced dance video at 720p and 30fps — the model handles coherent motion well past the 20-second limit that stops most AI video models. Wan Dancer (Wan-Dancer-14B) is Alibaba Tongyi Lab's music-to-dance AI model, released as open source under Apache-2.0. A global planning stage reads the full music track and lays out dance keyframes for the entire routine, then a local stage renders smooth frame-by-frame motion in sync with the beat. The hosted generator on this site is coming soon; until then, this page covers exactly how the model works and where it fits.
Music-Driven Portrait Dance Video — Wan Dancer
A music-driven portrait dance video model developed by Alibaba Tongyi Lab. Wan Dancer generates rhythm-synced dance videos from a single reference photo and a music track — no motion capture hardware and no reference dance video required. You start from one portrait photo of a person and an audio file, and it choreographs and renders that person dancing in time with the music at 720p and 30fps. Where most AI video models drift or fall apart after roughly 20 seconds, this model stays coherent for over a minute. It works by planning before it renders. A global stage reads the full music track and lays out the long-range dance as keyframes across the entire routine, then a local stage refines the motion between those keyframes frame by frame. That two-stage pipeline, built on the Wan image-to-video backbone, is what allows a still image and a song to become a full dance clip instead of a few seconds of movement.
- Music-Driven Choreography from a Single TrackThe model generates dance movement directly from your audio, so choreography follows the actual music rather than a generic loop. You supply the music track at the start and it aligns the motion to the rhythm and tempo. A text prompt guides the dance style and mood on top of the audio input.
- Minute-Scale Temporal CoherenceThe global-to-local pipeline holds the dance structure past the roughly 20-second point where most diffusion models break down. It produces dances that run over a minute — long enough for a full chorus or hook — while keeping the subject's identity consistent from the first frame to the last.
- Single-Photo Identity Lock with Genre VersatilityOne portrait photo is sufficient. The model carries the reference person's face, hair, and outfit through the entire routine across five dance genres — Chinese classical, K-pop, street, tap, and Latin. The dancer stays recognizable from start to finish regardless of the style selected through the prompt.
How Wan Dancer Generates a Dance Video
Wan Dancer separates planning from rendering — the global stage reads the full music track and lays out choreography as keyframes, then the local stage refines the motion between them frame by frame. Three steps from portrait and audio to finished dance video at 720p and 30fps.
Why Choose Wan Dancer for AI Dance Video Generation
A music-to-dance AI video generator built for long, rhythm-synced, single-subject choreography at 720p and 30fps. This model handles coherent output well past the 20-second limit that stops most diffusion models and produces dance clips over a minute long across five distinct dance genres.
Music-Driven Dance Generation
The Wan Dancer generator creates dance movement directly from your audio track, so choreography follows the actual music instead of a generic loop. You supply the music upfront rather than syncing audio to a silent clip afterward. A text prompt guides the style and mood on top of the audio for precise artistic direction.
Minute-Scale Temporal Coherence
The Wan Dancer global-to-local pipeline holds dance structure past the roughly 20-second point where most AI video diffusion models fall apart. It produces dances that run over a minute — long enough for a full chorus or song section — with consistent subject identity throughout the sequence.
Single-Photo Identity Lock
One portrait photo is sufficient for a full dance video. The model carries the reference person's face, hair, and outfit through the entire routine, so the dancer stays recognizable from the first frame to the last without re-uploading the source image.
720p at 30fps Output
The model renders high-definition video at a smooth 30 frames per second, ready for short-form social clips and music-video cuts on TikTok, Reels, and Shorts. The output is directly usable without additional post-processing or upscaling.
Five Dance Genres Trained from a Single Model
Trained across Chinese classical, K-pop, street, tap, and Latin, the model can match a range of musical styles from a single reference image. You choose the style through the text prompt alongside the audio track. Open weights under Apache-2.0 support self-hosting and LoRA fine-tuning for custom choreography.
Wan Dancer — Frequently Asked Questions
Answers about the model's capabilities, music-to-dance video generation, the two-stage global-to-local pipeline, minute-scale coherence, single-photo identity lock, the five supported dance genres, open-source availability under Apache-2.0, and differences from other AI video models.
What is Wan Dancer?
Wan Dancer (Wan-Dancer-14B) is an open-source music-to-dance AI video model from Alibaba's Tongyi Lab, released under the Apache-2.0 license. It takes one portrait photo of a person and a music track, then generates a video of that person dancing in time with the music at 720p and 30fps. It uses a two-stage pipeline where a global stage reads the full audio and plans the dance as keyframes, then a local stage refines the motion between them frame by frame.
How is it different from image-to-video models?
Wan Dancer is driven by music rather than a reference video. It choreographs a dance from an audio track and a single portrait photo, producing beat-synced movement at 720p and 30fps that stays coherent past the 20-second limit. Most image-to-video models work from a source video clip and handle character animation or replacement. Wan Dancer and Wan Animate are sibling models from the same lab but solve different tasks.
How long can a dance video be?
Wan Dancer is built for minute-scale generation and stays coherent well past the roughly 20-second point where most diffusion video models break down. Project demonstrations run over two minutes, and the open-source build targets shorter music inputs for practical use. The planning-first design allows dance clips long enough for a full chorus or song hook.
What do I need to use it?
You need a single reference image of a person — a vertical full-body shot works best — plus a music or audio file and a short text prompt naming the dance style. Clean low-noise audio and a clearly defined subject produce the steadiest dance results. A hosted generator on this site is coming soon.
What dance styles are supported?
The model was trained across five genres: Chinese classical, K-pop, street, tap, and Latin. You choose the style through your text prompt alongside the music track. The single-photo identity lock keeps the dancer's appearance consistent regardless of which genre is selected.
Is the model open source?
Yes. Wan-Dancer-14B is released under the Apache-2.0 license on Hugging Face and ModelScope, with inference code, ComfyUI integration, and LoRA fine-tuning support for custom choreography. You can self-host the model and use outputs commercially. A hosted generator on this site is coming soon for direct use.
Start Creating with Wan Dancer
Turn a single portrait photo and a music track into a beat-synced dance video. The two-stage pipeline — global choreography planning followed by local frame-by-frame refinement — delivers coherent dance output well past the 20-second mark at 720p and 30fps. The model is trained across five dance genres from a single portrait, with music-driven choreography that follows the actual rhythm, tempo, and mood of your audio track. No motion capture and no reference dance video required.