ByteDance / seedance-pro-v1.5-uncensored

## Overview Seedance Pro v1.5 Uncensored is a text-to-video model from ByteDance’s Seedance 1.5 Pro line. It generates video and optional audio in one pass, so speech and motion align. Use it for short cinematic clips with dialogue, foley, and music. ## What you can build - Dialogue scenes with visible lip-sync and matching speech audio - Short ad concepts with product shots, camera moves, and sound design - Social clips for 9:16 feeds with built-in ambience and music cues - Animatics for story beats, including mood, lighting, and sound notes - Multilingual versions of the same scene with localized dialogue ## Inputs - `prompt` (string, required): Describe the scene, actions, camera, and any desired audio. - `duration` (integer, default: 5): Clip length in seconds. This endpoint is commonly used for 5s and 10s runs. - `resolution` (string, default: "720p"): One of `480p`, `720p`, `1080p`. - `aspectRatio` (string, default: "16:9"): One of `21:9`, `16:9`, `4:3`, `1:1`, `3:4`, `9:16`. - `generateAudio` (boolean, default: true): When true, the model generates synchronized audio. - `cameraFixed` (boolean, default: false): When true, the camera stays locked. - `seed` (integer, default: -1): Set a fixed seed for more repeatable results. Use `-1` for random. ## Outputs - An MP4 video. - Audio is embedded in the MP4 when `generateAudio` is true. - The model targets a cinematic frame rate of 24 fps. ## Recommended settings - Fast iteration (preview cuts) - `resolution`: `480p` - `duration`: `5` - `generateAudio`: `false` - `cameraFixed`: `true` - Final dialogue clip (best lip-sync payoff) - `resolution`: `1080p` - `duration`: `10` - `generateAudio`: `true` - `cameraFixed`: `true` - Prompt: include exact quoted dialogue plus emotion and pacing. - Cinematic movement (when you want camera motion) - `resolution`: `720p` - `duration`: `5` - `cameraFixed`: `false` - Prompt: specify moves like pan, tilt, push-in, or orbit. ## Limitations - Clip length is short. Most workflows use 5s or 10s segments. - High-intensity action can reduce temporal stability. Fast limbs and complex collisions may jitter. - Lip-sync works best with clear, face-forward framing. Wide shots often hide mouth detail. - Audio is scene-driven. It may not match a precise script timing without careful prompting. ## Safety & compliance - This model can generate realistic people and voices. Don’t use it for impersonation or deception. - Don’t generate non-consensual sexual content, exploitative content, or illegal content. - Avoid copyrighted characters, logos, and brand assets unless you have rights. - Add your own moderation and access controls when you ship user-facing features.