Seedance 2.0 Fast R2V (Reference-to-Video)

Overview

Seedance 2.0 Fast R2V runs Dreamina Seedance 2.0 fast, a lighter model in ByteDance's Seedance 2.0 series, and needs at least one source video. ByteDance says it supports largely the same features as Seedance 2.0 and points to the full model when the highest generation quality is the goal. On this page the visible difference is resolution, capped at 720p.

One to three source clips go in, and the model returns a new shot that inherits their movement, camera path or style, with sound generated alongside the picture.

Choosing a variant: Fast R2V fits video watched at 720p or below, such as social feeds and in-app clips. For sharper masters or top quality, use Seedance 2.0 R2V. Without a source clip, Seedance 1.5 Pro generates from text or frames with sound, and Seedance 1.0 Pro does so silently.

What you can build

• Vertical 9:16 versions of the action in a horizontal clip, for short-form feeds.
• Spoken clips whose voice timbre comes from a reference audio file.
• Edits to an existing clip, such as removing an object or adding one at a set time and place.

Inputs

• 1 to 3 source videos in MP4 (required), each 2 seconds or longer. Wiro takes up to 15 seconds of footage across them in upload order, trimming or skipping whatever runs past that.
• An optional prompt that points at assets as [Video 1], [Image 2] and so on, numbered by upload order within each type.
• Up to 9 reference images. The first and last frame slots are not used on this page, because ByteDance keeps frame mode apart from runs with source videos.
• Up to 3 audio references in WAV or MP3, which Wiro requires to be paired with an image.
• Resolution 480p or 720p (default). Aspect ratio adaptive (default), 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16. Duration from 4 to 15 seconds, 5 by default.
• Switches for generated audio (on), watermark (off) and prompt enhancement (on), which rewrites your text into a brand-free form to cut content-filter failures. A seed of 0 is random.

Outputs

Output is 8-bit MP4 video at 24 fps. Generated audio is mono and follows the picture with effects, ambience, music or speech.

Recommended settings

• Use ByteDance's advanced prompt formula: subject, action details, scene, lighting and color, camera movement, visual style, image quality and constraints.
• For edits, describe what changes and repeat what must stay. Unmentioned parts are kept by default.
• For extension, write "Extend [Video 1] forward" (or backward) and then what happens. With two or three clips, describe each transition.
• To control the opening or closing shot, write in the prompt which reference image should be the first or last frame. ByteDance does not promise an exact match to that image.

Limitations

• The seed field gives no repeatability guarantee, as ByteDance lists seed support for Seedance 1.x only.
• Effects written only as text, such as digits counting down, may not follow exact logic. ByteDance suggests a reference video of the effect.

Safety & compliance

Use a recording of a real person's voice as an audio reference only if that speaker has agreed to it. For the 2.0 models, ByteDance blocks uploads containing real human faces, so such files may fail.