Seedance 2.0 Multimodal AI Video Generator

Create audio-synchronized videos from text, images, videos, and audio references with director-level control over motion, lighting, camera movement, and sound.

Create with Seedance 2.0

Model
Unified multimodal audio-video generation with text, image, video, and audio inputs.
Reference Mode
Image References (0/9)
Video References (0/3, max 15s total)
Audio References (0/3, max 15s total)
Creative Direction
174/500
Output Settings

Model: Seedance 2.0

Creative Direction: A pair of figure skaters perform a difficult lift and synchronized landing on reflective ice, with precise body motion, natural crowd ambience, and cinematic camera movement.

7/10/2026, 8:32:05 PM

Explore Seedance 2.0 multimodal video creations

Seedance 2.0 Motion

by AIArtist

video

Seedance 2.0 Character Reference

by AIArtist

video

Seedance 2.0 Cinematic Scene

by AIArtist

video

Seedance 2.0 Audio-Visual

by AIArtist

video

Seedance 2.0 FAQ - Unified Multimodal Audio-Video Generation

Seedance 2.0 is ByteDance's unified multimodal video generation model for creating audio-synchronized videos from text, images, videos, and audio references. It is designed for high-quality creative direction, combining visual generation, motion control, camera language, lighting, and native audio-video output in one workflow.
Seedance 2.0 supports text prompts plus optional reference media. You can guide a generation with image references, video references, and audio references, making it useful for character consistency, product videos, scene continuity, style matching, and sound-driven creative direction.
All-Round Reference mode lets you combine multiple types of creative references in one generation. In this interface, you can add up to 9 image references, 3 video references, and 3 audio references, then describe how Seedance 2.0 should use them for subject identity, visual style, movement, atmosphere, and sound.
First and Last Frames mode is designed for more controlled visual transitions. You can provide 1 or 2 image references as start and end frames, then use your prompt to describe how the scene should move, transform, or connect between those frames.
Seedance 2.0 is built for synchronized audio-video creation. When native audio-video generation is enabled, the model can create sound that matches the visual action, such as ambience, foley, movement sounds, product sounds, dialogue-like timing, or ASMR-style audio cues depending on the prompt and references.
This Seedance 2.0 tool supports 4, 5, 8, 10, and 15 second outputs with 480p and 720p resolution options. It also supports multiple aspect ratios including 16:9, 9:16, 4:3, 1:1, 3:4, 21:9, and adaptive ratio, so you can create content for widescreen, mobile, square, and cinematic formats.
Seedance 2.0 is suited for cinematic shorts, product films, social media clips, character-driven scenes, music and performance videos, ASMR-style videos, marketing content, motion experiments, and reference-based creative work where visual continuity and synchronized sound matter.
Write prompts like creative direction for a film shot. Include the subject, story beat, motion, camera movement, lighting, mood, sound, pacing, and any relationship to your uploaded references. For example, describe whether the camera pushes in, tracks behind the subject, cuts between shots, follows a product rotation, or syncs sound effects to specific actions.
Seedance 2.0 uses credits based on output resolution, duration, and whether video or audio references are used. In this tool, 480p starts at 20 credits per second and 720p starts at 40 credits per second. Generations with video or audio references may require additional credits because they involve more multimodal processing.