Seedance 2.0 Mini

Seedance 2.0 Mini — Multimodal AI Video Generation (Text-to-Video & Image-to-Video)

What is Seedance 2.0 Mini?

Seedance 2.0 Mini is ByteDance's lightweight, efficiency-optimized tier of the Seedance 2.0 video family. It is a distilled version of the flagship model: it keeps the same unified multimodal architecture but runs roughly twice as fast as Seedance 2.0 Fast, making it built for high-volume, iteration-heavy video generation. With one API call you can run text-to-video, image-to-video, or reference-to-video — turning a written prompt, a first-frame image, or attached reference media into a short clip with motion, camera movement, lighting, and optional synchronized audio.

Key Features

• •Multimodal inputs: Generate from text alone, animate a still via first_frame_url, or steer output with up to 9 reference images, 3 reference videos, and 3 reference audio clips.
• •@-style reference tags: Cite uploaded assets as image 1, video 1, audio 1 in your prompt to lock character identity, transfer motion, and align sound.
• •Native audio co-generation: Enable generate_audio to produce dialogue, sound effects, ambient sound, and music in the same pass as the visuals.
• •Multi-shot direction: Number shots (Shot 1, Shot 2) and use plain-language camera language — dolly, pan, orbit, tracking — within a single generation.
• •Flexible output: Durations of 4–15 seconds; aspect ratios 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, and adaptive; resolutions from 480p to 720p for fast iteration.

Best Use Cases

Mini is the default choice for short-form vertical video (TikTok, Reels, Shorts), e-commerce and product demos, marketing ad variations, anime and illustration work, and UGC. Its speed makes it ideal for testing many prompt, framing, and motion variants quickly. The recommended professional workflow is to iterate on Mini, then re-render only your final approved shots on standard Seedance 2.0 when peak fidelity matters.

Prompt Tips and Output Quality

Write a structured shot list rather than prose: subject + action + environment + camera + style, plus an optional sound line. Seedance does not use negative prompts — use positive constraints like "stable face, consistent outfit, natural anatomy." Keep one main action verb per shot, limit scenes to one or two characters, and prefer gentle motion words (slow, smooth, continuous). Always close with a global style line for consistent look. Note that real human faces are blocked across all Seedance tiers.

How to Use Seedance 2.0 Mini

Seedance 2.0 Mini is ByteDance's fast, lightweight tier of the Seedance 2.0 family. It generates short video from text, a first-frame image, or multimodal references, with optional synchronized audio. Use it to iterate quickly and at volume, then graduate final shots to standard Seedance 2.0 when you need maximum fidelity.

Text-to-Video

Send a prompt only. Write a structured shot list rather than a single sentence: subject + action + environment + camera + style. Specify camera language explicitly — slow dolly in, pan, orbit, tracking shot, static camera. Close with a global style line (for example, cinematic lighting, 35mm, film grain) so the look stays consistent.

Image-to-Video

Pass a first_frame_url to animate a still outward from that frame. Add a last_frame_url (requires a first frame) to control the transition between two frames. Note: images containing real human faces are blocked, and first_frame_url cannot be combined with reference_images.

Reference-to-Video

Attach up to 9 reference_images, 3 reference_videos, and 3 reference_audios, then cite them in the prompt as image 1, video 1, audio 1. Use images for character or style consistency, videos for motion/camera transfer, and audio for rhythm or voice. Keep reference videos short and clean for more precise results.

Multi-Shot Sequences

Number your shots (Shot 1, Shot 2) or use timestamps to structure a narrative. Keep the subject noun consistent across shots, one main verb per shot, and one or two characters maximum to avoid identity drift. There are no negative prompts — use positive constraints such as "stable face, consistent outfit, natural anatomy."

Recommended Settings

• •Drafts / fast iteration: resolution 480p, duration 4–5, generate_audio false.
• •Final short-form: resolution 720p, duration 5–10, aspect_ratio 9:16 for social or 16:9 for landscape.
• •Cinematic narrative: duration 12–15, generate_audio true, add a sound-design line in the prompt.
• •Reproducibility: set a fixed seed to iterate on the same scene; use -1 for fresh variations.
• •Chaining clips: enable return_last_frame, then pass the returned frame as the next first_frame_url.

Is Seedance 2.0 Mini faster than Seedance 2.0 Fast?

Yes — it is reported to run roughly twice as fast at comparable or better quality, which is why it largely supersedes the older Fast tier for drafts and iteration.

What is the difference between Mini and standard Seedance 2.0?

Mini is the lightweight iteration tier optimized for speed and volume; standard Seedance 2.0 is the highest-fidelity tier for final, polished, higher-resolution deliverables.

Does Seedance 2.0 Mini support text-to-video and image-to-video?

Yes. It supports text-to-video, image-to-video (via first_frame_url), and reference-to-video with image, video, and audio references.

Can it generate real human faces?

No. Face-blocking content policy applies across all Seedance tiers, so images with real human faces are rejected.

Does Mini generate audio?

Yes. Turn on generate_audio to co-generate synchronized dialogue, sound effects, ambient sound, and music alongside the video.

How long can the videos be?

Clips range from 4 to 15 seconds per generation, selectable via the duration parameter.