Seedance 2.0 Mini Reference to Video

Compact Seedance 2.0 variant for multimodal reference-to-video. Up to 9 reference images, 3 reference videos, 3 audio clips with native audio output.

"Beautiful fusion of these two scenes. Mills stand against a rugged coastline, their large wooden wheels turned by the relentless surge of tidal waves combined with a field of wildflowers bathed in soft sunlight transitions into where monarch butterflies take flight.."

Image 1

Image 2

Generated Result

Generated

Describe your scene and generate a video in seconds

8,500+ videos generated this month

📄 About Seedance 2.0 Mini Reference to Video

Seedance 2.0 Mini Reference to Video is the multimodal entry in the Seedance 2.0 Mini family. Drop up to 9 reference images, 3 reference videos, and 3 reference audio clips, then describe the scene by tagging your references as @Image1, @Video1, @Audio1 etc. The model composes these inputs into a 4–15 second cinematic video with synchronized native audio. Mini is the right pick when your concept needs more than a single image to communicate — character + environment + motion style, or product + audio brand cue — but you don't want the full Pro tier's cost. The same intuitive @reference syntax used in Pro applies here, so prompts portable across tiers. For maximum quality and longest durations, use the standard <a href="/model/seedance-2-0-reference-to-video">Seedance 2.0 Reference to Video</a>. For cheapest iteration, use <a href="/model/seedance-2-0-fast-reference-to-video">Seedance 2.0 Fast Reference to Video</a>.

✨ Key Features

Up to 9 reference images, 3 reference videos, 3 reference audio inputs.

Native audio output synced to the generated video.

@Image1 / @Video1 / @Audio1 prompt syntax for precise control.

Seven aspect ratios and 480p/720p resolution.

Cinematic physics and motion continuity.

💡 Use Cases

⚡Character + environment composition for branded shorts.

⚡Ad creative built from product image + reference music + reference motion video.

⚡Editorial photo + audio narration motion pieces.

⚡Concept-board reference videos for pitch decks.

⚡Music video previz from mood-board references.

🎯 Best For

🎯 Creators and marketing teams who need to compose video from multiple reference modalities at balanced cost.

👍 Pros

✓Multimodal reference inputs in one model

✓@reference tagging keeps prompts deterministic

✓Native audio in one pass

✓Balanced cost between Fast and Pro tiers

⚠️ Considerations

△Max 15 seconds

△Max 720p output

△Quality depends on reference clarity and audio cleanliness

📚 How to Use Seedance 2.0 Mini Reference to Video

Upload up to 9 reference images and optionally up to 3 videos and 3 audios.

Write a prompt that tags references explicitly: @Image1, @Video1, @Audio1.

Pick aspect ratio + duration + resolution.

Keep audio on; the model will mix synced audio output.

Generate, review, refine prompt or swap references for variations.

Ready to try Seedance 2.0 Mini Reference to Video?

Get 10 free credits — no credit card required

Start Free →

Frequently Asked Questions

Each uploaded reference gets an index (Image1, Image2, ..., Video1, Audio1, ...). In your prompt, use @ImageN/@VideoN/@AudioN to point at a specific reference. Example: '@Image1 walks through @Image2 while @Audio1 plays.'

No, but every uploaded reference will influence the output to some degree. For sharper control, only upload the references you actually reference in the prompt.

Audio references seed the model's audio output — mood, tempo, ambient texture. Note: reference audio cannot be used without at least one image or video reference.

15 seconds. For longer narratives, generate clips sequentially and stitch in post.

⚖️ How Seedance 2.0 Mini Reference to Video Compares

Seedance 2.0 Mini Reference to Video balances cost and multimodal capability between Seedance 2.0 Fast Reference to Video and the full Seedance 2.0 Reference to Video. Pick Mini for production-grade quality at a controlled cost; Fast for cheapest iteration; Pro for max resolution and duration. For competing reference-driven models also see Google Veo 3.1 Reference-to-Video and Kling O1 Reference to Video.