JustDance focuses on turning people, pets, and illustrated characters into short dance or motion videos. Upload one image and describe the movement and camera, with no filming or editing required.
Each example shows a reference or source image, motion prompt, and corresponding video. Check subject consistency, motion continuity, and background behavior before deciding whether the workflow fits your image.
A cute cartoon toddler in a colorful traditional outfit performs a light dance in a warm, tidy bedroom, taking small side steps and waving both hands. Keep the face, outfit, and background stable. Fixed full-body camera, vertical 9:16.
A golden retriever happily dances upright, moving its front paws to the beat. Keep the face and fur clear, use a simple stable background, and frame the full body with a fixed camera. Playful mood, vertical 9:16.
A girl group performs clear, synchronized K-Pop choreography. Keep each person's appearance and outfit consistent, maintain stable stage lighting, and use a medium camera shot that follows the movement.
Choose it when you already have a picture of the exact subject you want to see move. If your need is different, the last point tells you where to go instead.
Because it starts from your image, the face, outfit, or fur it produces stays much closer to the original than describing the same subject in words ever could.
Turn a single portrait, pet photo, or product-style shot into motion without setting up a shoot, a performer, or an editing timeline.
Drawn characters, mascots, and stylized avatars work well as long as the subject is clearly outlined and not heavily obstructed.
With only an idea and no picture, start from Text to Video. If you already have a video and want a new visual style, use Video to Video instead.
Your first generation
Check whether the image is suitable, describe one clear movement, then refine the result by problem type. This makes it easier to identify whether the image, motion, camera, or model needs to change instead of regenerating without direction.
Confirm that the subject is clear, limbs are not cropped, lighting is even, and the background is not distracting. Problems in the source image usually carry into the video.
Write the subject and specific movement first, then add the setting, stability requirement, and camera. Validate the subject with a fixed camera before adding turns, jumps, camera moves, or scene changes.
When the result is wrong, identify the failure first. Change only one variable—the image, motion, camera, or model—before generating again.
Create a photorealistic source-image comparison for an AI image-to-video product guide, landscape 16:9, clean split-screen composition showing two Asian female street dancers with a similar visual style, both wearing simple dark gray sportswear and white sneakers. Left side, suitable source image: the dancer fills most of the frame, relaxed front three-quarter standing pose, entire head, both hands, and both feet visible, clear silhouette, soft even daylight, plain warm-gray studio background, sharp focus, realistic skin and natural body proportions. Right side, unsuitable source image: the dancer is small in the frame, feet cropped, one hand obscured, strong backlight, visible motion blur, crowded street background with several people and visual clutter. Professional product-education photography, realistic camera look, neutral color palette, obvious visual contrast between good and bad source quality. No text, labels, arrows, logos, watermarks, UI borders, or brand elements anywhere in the image.
Using one clear full-body reference image as the first frame, generate an 8-second photorealistic vertical 9:16 video. The subject is a 25-year-old Asian female street dancer wearing a loose dark gray tracksuit and white sneakers in a simple warm-gray studio. She performs one low-complexity dance sequence: take a small step to the left while naturally swinging both arms, step to the right while raising one hand, then return to a centered standing pose. Keep the rhythm clear, movement moderate, balance stable, and transitions smooth. Use a fixed full-body camera for the entire clip; keep the complete person centered and visible. No turns, jumps, fast head movement, zoom, pan, camera shake, or cuts. Strictly preserve facial identity, hairstyle, clothing colors, body proportions, and shoes. Keep fingers anatomically correct, feet stable, and the background, floor, and lighting completely static. Realistic camera footage, soft even lighting, 24 fps, natural motion blur, loop-friendly ending. No captions, logos, watermarks, extra people, lens distortion, deformed limbs, or background drift.
Create an independent educational AI video quality-review contact sheet, landscape 16:9, with four equal cinematic panels. The panels demonstrate how to inspect facial identity, hand structure, foot contact, and background stability; do not imply that they are consecutive frames extracted from a real video. Panel one: a sharp facial close-up. Panel two: a medium shot of both hands and arms in motion. Panel three: a lower-body shot of both feet stepping. Panel four: a centered full-body wide shot. Photorealistic professional video-review reference, sharp but natural. No text, numbers, labels, arrows, logos, watermarks, UI controls, extra people, or malformed limbs.
A useful prompt does not need to be long. It needs five kinds of information the model can act on.
Example you can edit
A golden retriever happily dances upright, moving its front paws from side to side with the beat in a bright, tidy living room. Keep the face, fur, and background stable. Fixed full-body camera, vertical 9:16.
These four checks matter more than adding more prompt text.
Ready to try
Replace the image first
Troubleshooting
Do not change every setting at once. Start with the most likely cause of the visible problem, then change only one variable before generating again.
Likely cause
The source is unclear, or the motion includes large turns and extreme movement.
Change first
Use a clear front-facing image, reduce motion intensity, and ask the model to keep the face and outfit stable.
The image guides identity and the prompt guides motion, but the output is regenerated frame by frame. Knowing these boundaries saves you from regenerating without a reason.
The photo is a strong reference, not an exact lock. A clear, front-facing, well-lit subject keeps identity much closer; heavy angles or blur let it drift.
If hands or feet are cut off or covered in the source, the model has to guess them, which is where most deformation appears. Start from a fuller view instead.
Generation varies run to run. When comparing settings, keep the image, prompt, duration, and resolution fixed and change one thing at a time.
Big turns, multiple people, outfit changes, and strong camera moves are each manageable alone but harder to keep stable when requested together.
These scenario pages carry source-image guidance and examples tuned to one subject, so you get more relevant advice than a general workflow can give.
Baby dance
Age-appropriate motion guidance, safer movement examples, and guardian-permission reminders for creating and sharing baby dance videos.
Open the baby dance generatorPet dance
Photo tips and movement examples matched to an animal's body, so a dog, cat, or other pet stays recognizable while it moves.
Open the pet dance generatorNot sure which to pick? Start with Seedance 2.0 at 8 seconds—it fits most first generations. Move to Hailuo 2.3 when you need up to 1080p or a more stylized look.
More durations and ratios
Offers 7 aspect ratios and 4, 8, or 12-second durations, giving you more output choices for different platforms and content lengths.
Up to 1080p
Offers output up to 1080p with 6 or 10-second durations for comparing higher-resolution or more stylized results.
Answers specific to generating from an image—what to upload, how consistent it stays, and where an end frame helps.
Register to receive trial credits and start generating. Your current credit balance and generation cost are shown in your account and generator.
