Seedance 2.0 Mini Reference to Video: Images to 21:9 Clips

Last verified: November 3, 2026

Seedance 2.0 Mini works best for short clips that have to match material you already have, like catalogue shots of a product, stills of a character or photos of a location. It can frame the result as a wide 21:9 letterbox just as easily as a vertical or square shot. Seedance 2.0 Mini reference-to-video reads a set of images together with your prompt. It can also take a reference video and an audio track in the same request, and it renders the soundtrack in the same pass as the picture.

In practice, the references set the look and the prompt sets the action. That split works well when the visual identity is decided before the shot is: a product launch built from catalogue images, a character spot made from a few stills of the same person, or a location teaser timed to a music cue. The prompt doesn't have to describe every detail again, so it can spend its words on movement, camera and mood.

On Vidofy, the images are the only required part. A reference clip and an audio file are optional, and they're only worth adding when the movement or sound has to follow something specific. Clip length, resolution, frame shape and the audio switch sit below the prompt, and the steps further down explain what to pick for each one.

Explore more reference to video

Reference To Video Use in First To Last Frame Seedance 2.5 Kling 4.0 Vidu Q4 Wan 3.0 Prime MiniMax H3 Max MiniMax H3
Capability Snapshot

Inputs, Clip Lengths and Output at a Glance

What this generator accepts and what one run gives you, all in one place.

Input

Up to 9 images (JPG, JPEG, PNG or WEBP, up to 10 MB each) + up to 3 videos (MP4 or MOV, up to 50 MB each) (optional) + up to 3 audio files (MP3 or WAV, up to 15 MB each) (optional) + a text prompt of up to 2,048 characters

Video Duration

4 to 15 seconds

Resolution

480P or 720P

Aspect Ratio

Auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16

Audio

Switch on or off, on by default

Output

1 video per run

How to Get Better Results from Seedance 2.0 Mini

Draft at the Lower Resolution, Save the Higher One for the Final

Seedance 2.0 Mini renders at two resolutions, so treat them as a draft and a master. While you're still swapping reference images and rewording the prompt, leave Resolution on the lower setting. Motion, framing and whether the references come through correctly are all easy to judge there. Once a version works, change only Resolution to the higher setting and generate again with the same uploads, prompt, Video Duration and Aspect Ratio. With everything else fixed, the final render is the closest comparison you can get to the draft you approved.

Fit the Action to the Clip Length

Mini renders at 24 frames per second across its whole duration range, so plan the prompt around the time you set in Video Duration. The default length leaves room for one clear action or one camera move. When you need a sequence, such as a turn, a reveal and a closing pose, raise Video Duration and write the beats in the order they should happen. Cramming three events into the shortest setting usually means some of them get rushed or left out.

How to Turn Reference Images into a Clip with Seedance 2.0 Mini

Six steps take you from uploading references to a downloaded clip.

1

Upload reference images in "Add Image"

Start in "Add Image", the only upload the form requires. Choose stills that each bring something different, like the subject, the setting, a product detail or the colour mood, rather than several near-identical shots, so every image gives the prompt a separate reference to work with. Use clean, well-lit originals rather than screenshots or heavily compressed copies.

2

Add optional clips in "Add Video" or "Add Audio"

If the new shot should follow existing footage, add it in "Add Video". If a track or voice should shape the sound, add it in "Add Audio". Both are optional, so skip this step when the stills and prompt already say enough. Trim reference clips to the part you actually want followed, since length limits apply (see the upload FAQ below).

3

Describe the shot in "Prompt"

Write the action, camera and mood in "Prompt", which holds up to 2,048 characters. Refer to your uploads by what they show, such as the woman in the denim jacket, the rooftop café or the drum loop, so each reference has a clear role. If you only have a rough idea, the "Enhance Prompt" button under the box suggests fuller versions. Your text only changes when you pick one, so you can compare suggestions without losing your draft.

4

Set "Video Duration", "Resolution" and "Aspect Ratio"

"Video Duration" runs from 4 to 15 seconds in one-second steps and starts at 4. "Resolution" offers 480P and 720P, with 480P selected by default. "Aspect Ratio" offers Auto, 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16 and starts on 1:1, so change it before you generate if you want anything other than a square. Pick 21:9 for a cinematic letterbox, 16:9 for standard widescreen or 9:16 for vertical social feeds.

5

Check "Generate Audio" and click "Generate"

"Generate Audio" is on by default, so your clip comes with sound. Switch it off if you'll add music or a voice-over in your own edit and want a clean silent clip to work with. Then click "Generate" to start the run.

6

Download your clip

Each run returns one video. Watch it all the way through before you download. If a reference was misread or the action ran out of time, adjust the prompt or settings and generate again; that's usually quicker than trying to fix it afterwards.

Reference Images or a Fixed First Frame: Which Mode to Use

When your stills should guide the clip, and when a single picture should set its frames.

What your images do Here the uploads act as references for the model, and the prompt decides the shot. The model's documentation lists this as a separate function from frame-based image-to-video. In Image To Video, Seedance 2.0 Mini is documented to work from a first frame, or from a first and last frame, so your image sets where the clip begins. Pick Image To Video when the clip has to open on one exact picture. Stay with reference mode when your stills are material the model should build a new shot from.
How many images feed one clip Reference generation can draw on a whole set of images in a single run, so a subject, a setting and a detail shot can all inform the same clip. Image-to-video takes one image as the first frame, or two images when a last frame is added. Choose image-to-video when you have one keyframe or a start-and-end pair. Choose reference mode when the clip has to combine several pieces of visual material.
Stay with reference mode when several stills should inform one shot. Switch to Image To Video when one picture, or a start-and-end pair, has to set the frames themselves.

Seedance 2.0 Mini Problems and How to Fix Them

The issues that come up most when building clips from references, and what to change.

I only want to use a music track as the reference, with no pictures.

The form needs at least one still in Add Image for every run, so audio can only be added alongside an image, never on its own. Add a still that sets the look of the scene; one location or mood image is enough to meet the requirement. Keep your track in Add Audio, describe in the prompt how the picture should move with the music, and generate again.

My reference video won't upload.

Each reference clip is checked for format, file size and length, and all the clips together share one length budget. That means two clips can fail together even when each would pass on its own. The exact limits are in the upload FAQ below. Trim each clip down to the part that shows the movement or framing you want, export it in one of the accepted formats, then add it again and generate.

The form won't take my reference image.

Stills are checked for format, file size and minimum pixel size. Phone photos saved as HEIC, animated GIFs, tiny thumbnails and very large exports are the usual problems. Re-save the image in one of the accepted formats listed in the upload FAQ. Shrink files that are too large, and replace anything below the minimum size with a larger original rather than upscaling it. Then upload it again.

My video came out square, but I wanted widescreen.

The frame shape comes from Aspect Ratio, which starts on the square option. Before you click Generate, switch it to 21:9 for a cinematic letterbox or 16:9 for standard widescreen. Rewrite any framing notes in the prompt for the wider frame, then run it again.

The clip ends before the action finishes.

Video Duration starts at its shortest length, which leaves room for about one gesture or one camera move. Either raise Video Duration toward the top of its range, or cut the prompt down to the one beat that matters most. Then generate again; length is set before the run, so you can't extend a finished clip from this form.

Frequently Asked Questions

How many reference images, videos and audio files can Seedance 2.0 Mini use?

Add Image takes 1 to 9 images in JPG, JPEG, PNG or WEBP format. Each can be up to 10 MB and must be at least 300×300 pixels, and at least one image is required. Add Video is optional and takes up to 3 MP4 or MOV clips of up to 50 MB each; no clip can run over 15 seconds, and all clips together are capped at 15 seconds. Add Audio is optional too: up to 3 MP3 or WAV files of up to 15 MB each, with the same 15-second limit per file and 15 seconds combined.

Does Seedance 2.0 Mini generate sound along with the video?

Yes. Seedance 2.0 Mini is documented as an audio-and-video co-generation model, so the sound is made in the same run as the picture. Here, the Generate Audio switch decides whether you get it: leave it on for a clip with sound, or turn it off when the soundtrack will come from your own edit. A file in Add Audio does a different job; it's a reference for the model to work from.

What makes one Seedance 2.0 Mini generation cost more than another?

The credit cost of a run is calculated from the options you choose and shown above the Generate button once you're signed in. It changes with Video Duration, with Resolution, with whether you upload a reference video, and with how long the uploaded videos are. Settle those choices first, then check the figure before you click Generate.

Can I generate a clip without uploading an image?

No. Add Image is the only required upload here, so every run needs a still, even if you also add video or audio. If you want to start from a written description alone, the Text To Video page runs the same model from a prompt.