Turn a written shot or a set of visual and audio references into a controlled AI video with Seedance 2.0 Mini. Choose 480p or 720p, set a 4–15 second duration, and guide identity, movement, composition, and pacing with the input mode that fits your project.
Seedance 2.0 Mini Guide
A practical guide to supported inputs, workflow selection, output controls, and prompt strategy.
Seedance 2.0 Mini is a ByteDance AI video model built for text, frame-guided, and multimodal reference generation.
It supports 480p and 720p output, durations from 4 to 15 seconds, and a broad set of square, portrait, landscape, ultrawide, and adaptive aspect ratios.
Use pure text-to-video when the scene should be invented from a prompt. Use a starting frame to animate an existing visual, or first and last frames when both endpoints matter.
Use multimodal reference mode when images, videos, or audio should guide the result. It accepts up to 9 images, 3 videos, and 3 audio files, but audio must accompany visual reference media.
Describe one primary subject and action, then state the setting, framing, camera movement, visual style, and pacing. Keep instructions consistent with the selected references.
Review the first result for identity, continuity, and timing. Remove conflicting references or ambiguous actions before generating a refined variation.
Primary reference
Start from a prompt, animate a first frame, define first and last frames, or guide the clip with a multimodal reference set. Each workflow keeps inputs aligned with the intended creative task.
Try Seedance 2.0 MiniUse up to 9 images, 3 videos, and 3 audio files as references to guide subjects, visual language, movement, rhythm, and sound cues. Audio must be paired with visual reference media.
Try Seedance 2.0 MiniSet any supported duration from 4 to 15 seconds and choose 480p or 720p output. Multiple aspect ratios cover square, portrait, landscape, and ultrawide publishing formats.
Try Seedance 2.0 MiniFirst/last-frame control and multimodal reference mode are separate workflows. Keeping them separate reduces conflicting guidance, while web search is available only for pure text-to-video generation.
Try Seedance 2.0 MiniChoose the correct input workflow, direct one clear shot, and refine the generated result.
Use text-to-video for a new scene, image-to-video to animate a frame, first/last frames for endpoint control, or reference mode for a mixed image, video, and audio brief.
Write a focused prompt describing subject, action, camera, setting, and pacing. Choose an aspect ratio, 480p or 720p resolution, and a duration between 4 and 15 seconds.
Check identity, motion continuity, composition, and timing. Simplify conflicting instructions or replace weak references, then generate another variation and download the preferred result.
Seedance 2.0 Mini is a ByteDance AI video model for prompt-based, frame-guided, and multimodal reference generation with 480p or 720p output and durations from 4 to 15 seconds.
Build controlled AI video from a prompt, frames, or multimodal references.
Try Seedance 2.0 Mini