Seedance can take a still image and turn it into a moving video clip — the camera pans, subjects move, environments come alive. The results are impressive when the prompt and settings are right.
This guide covers the image-to-video workflow from start to finish: how to prepare your image, write the motion prompt, choose the right settings, and fix common problems.
Quick answer
Upload your source image, write a prompt describing the desired motion (not the image itself — it already knows what the image shows), set the motion intensity and duration, and generate. Best results come from high-quality source images with clear subjects and simple backgrounds.
- You have a still image you want to animate with realistic motion.
- You want to create video content from existing photography or artwork.
- You need short video clips for social media, ads, or presentations.
Preparing your source image
The quality of your source image determines the quality ceiling of your video. Use high-resolution images (at least 1024x1024) with clear subjects, good lighting, and simple backgrounds.
Avoid heavily compressed images, images with text overlays, or images with complex busy backgrounds. Seedance handles these poorly and introduces artifacts.
Writing the motion prompt
The prompt should describe motion, not the image content. Seedance already sees the image — telling it 'a woman standing in a field' wastes prompt space. Instead, describe what should move and how.
Good motion prompts are specific about direction, speed, and type of movement.
- Describe motion: 'camera slowly pans right', 'subject turns head toward camera'
- Specify speed: 'gentle breeze', 'quick zoom', 'slow dolly forward'
- Avoid contradictions: do not ask for both 'camera pans left' and 'subject walks right to left'
- Keep it focused: one or two motions work better than five competing movements
Choosing settings
Motion intensity controls how much movement appears. Low values give subtle, realistic motion. High values create dramatic movement but risk distortion.
Duration affects quality — shorter clips (3-5 seconds) are more consistent than longer ones. Start with 4-second clips and extend only if quality is acceptable.
Common problems and fixes
If the subject's face distorts during movement, reduce motion intensity and use a simpler motion direction. If the background warps, use a higher-quality source image with more background detail.
Flickering usually means the motion prompt is too complex. Simplify to a single motion direction.
Getting the best results
Generate 3-4 variations of each clip. Seedance produces different results each time, and the quality varies. Pick the best one and refine the prompt based on what worked.
For consistent quality, develop a template prompt structure that you know works well with your image style, then adapt it for each new image.
Worked example: animating a product photo
You have a product photo of a coffee cup on a wooden table. Prompt: 'Camera slowly dollies forward, steam rises gently from the cup, warm morning light shifts slightly.' Settings: motion intensity 40%, duration 4 seconds. Result: a professional-looking product video from a single photograph.
Common mistakes
- Describing the image content in the prompt instead of the desired motion.
- Setting motion intensity too high, causing distortion and artifacts.
- Using low-resolution or heavily compressed source images.
Step by step: turn one image into a clean video
- Pick a source image with clear focal depth. A portrait with soft background works better than a flat landscape.
- Crop to your target aspect ratio before uploading. 16:9 for YouTube, 9:16 for Shorts. Seedance will not recrop cleanly.
- Upload as the reference image. In Dreamina, drag into the "Reference" slot — not the prompt box.
- Write a motion prompt, not a scene prompt. "The woman slowly turns her head to the left" beats re-describing the whole image.
- Cap motion intensity at 40. Image-to-video is most stable below 50. Above 60, faces warp.
- Generate 5 seconds first. Longer clips drift further from the reference.
Troubleshooting table
| Symptom | Likely cause | Fix |
|---|---|---|
| Output looks nothing like the reference | Reference slot not used | Make sure the image went into the reference slot, not the prompt. |
| Face warps halfway through | Motion intensity too high | Drop to 30-35 for face-centric clips. |
| Background moves but subject stands still | Prompt described scene, not subject motion | Rewrite prompt to describe what the subject does. |
| Aspect ratio changes unexpectedly | Uploaded a different ratio | Pre-crop the image to match your target ratio. |
For aspect ratio and export rules, see resolution and export settings. For character consistency across multiple clips, see reference images for characters.
When to use something else
For writing better prompts overall, see better prompts for Seedance. For using the Dreamina platform, see Seedance in Dreamina.
Frequently asked questions
What should the prompt describe in image-to-video?
The motion, not the image. Seedance already sees the source image, so re-describing 'a woman standing in a field' wastes the prompt — say what should move and how, such as 'she slowly turns her head to the left'.
What makes a good source image?
High resolution (at least 1024x1024), a clear subject, good lighting and a relatively simple background. The source image sets the quality ceiling — a compressed or cluttered image limits how good the video can be.
What motion intensity should I use?
Cap it around 40 for most clips and 30-35 for anything face-centric. Image-to-video is most stable below 50; above 60, faces tend to warp.
How long should image-to-video clips be?
Start at 4-5 seconds. Shorter clips stay closer to the reference; longer ones drift further from it, so generate short and concatenate in an editor for longer sequences.
Why does my output look nothing like the source image?
The image probably went into the prompt box rather than the reference slot. In Dreamina, drop it into the 'Reference' slot, then write only the motion in the prompt.
Why does the background move but the subject stay still?
The prompt described the scene instead of the subject's action. Rewrite it to say what the subject does — 'the man raises his cup' — rather than re-describing what is in the frame.
Gemini Omni image-to-video workflow
If you are testing reference preservation across models, pair this Seedance guide with Gemini Omni Image to Video and keep the same reference asset, duration, and aspect ratio for a fair comparison.
Related guides on this site
These guides cover prompt writing, platform setup, and motion control for Seedance.