Seedance 2.0

Positioning Overview

Seedance 2.0 Mini is the lightweight line in ByteDance's Seedance 2.0 family, called through the unified HiAPI.ai interface, and supports text-to-video, image-to-video, and reference-material-to-video generation. It can accept up to 9 reference images, 3 reference videos, and 3 reference audio clips, using visuals, motion, and timbre as generation conditions to output 4–15 second short videos with audio at 480p or 720p. It is suitable for developers who need to produce social materials, advertising previews, and creative samples in batches within short cycles.

Core Highlights

• Three-way conditional input with reference images, videos, and audio — reference materials can simultaneously constrain visual style, motion, and audio, reducing the uncertainty of pure text descriptions.
• First and last frames control the start and end of the shot — by providing both first-frame and last-frame images, you can specify where the visuals begin and where they converge, making it easier to connect multiple shots.
• Optional native audio generation — by default, audio synchronized with the visuals is generated, but the audio channel can also be turned off to produce only what the visuals require.
• Parameterized duration and aspect ratio — duration is adjustable from 4–15 seconds and supports aspect ratios such as 1:1, 16:9, 9:16, and 21:9, adapting to both landscape and portrait distribution.

Capability Description

Three generation paths: text prompts, first-frame images, or reference materials can all serve as input starting points. Choose the generation method based on the materials you already have, with no need to switch interfaces for different paths.

Reference materials constrain visuals and motion: up to 9 reference images (combined with first and last frames), and up to 3 reference videos with a total duration of no more than 15 seconds. Use real materials instead of long text descriptions to lock in style and action.

Reference audio guides timbre: pass in up to 3 reference audio clips with a total duration of no more than 15 seconds, so the generated result is closer to the given samples in timbre and audio character.

Resolution and audio track tiers: 480p generation is faster, while 720p balances speed and image quality; whether audio is generated and the resolution together determine the billing tier, allowing combinations to be selected according to budget.

Applicable Scenarios

• Batch production of short videos for social platforms — operations staff use short prompts or product images to generate 9:16 vertical clips and publish them directly to feeds and short-video accounts.
• Advertising storyboard previews — creative teams use reference images and first/last frames to output 4–15 second preview clips, aligning shot rhythm and visual tone before formal shooting.
• Brand material style extension — designers upload brand reference images and short reference videos to generate a series of clips with consistent style for landing pages and display placements.
• Rapid prototyping of short films with sound effects — content creators enable audio generation and pair it with reference audio to produce audio-visually synchronized samples for review and secondary editing.

Limitations and Selection Advice

The maximum output duration per request is 15 seconds, and billing is based on output seconds, so long content needs to be split into multiple segments and then stitched together. First-frame and reference materials are mutually exclusive scenarios and cannot be mixed in the same request; reference materials are subject to total image count and total duration limits, so projects with many materials need to be processed in batches. Resolution supports only 480p and 720p; scenarios requiring high-resolution final outputs should use higher-specification lines within the family.