Seedance 1 Lite

Generate videos from text prompts or images. Supports 5s or 10s clips at 480p and 720p.

Model Overview

A video generation model that creates videos from text prompts and images, supporting both text-to-video and image-to-video generation.

Best At

Creating short video clips (5s or 10s) with good motion and visual consistency from text descriptions or by animating existing images. It excels at interpreting nuanced prompts for diverse visual styles and camera movements.

Limitations / Not Good At

While it supports resolutions up to 720p, the documentation mentions 1080p output for the "Seedance 1.0" version, which may be a discrepancy. Reference images cannot be used with 1080p resolution or first/last frame images. Actual generation quality can vary with prompt complexity and may not always be perfect with multi-shot consistency.

Ideal Use Cases

• Generating short illustrative clips for social media or blog posts.
• Creating animated sequences from still images for presentations.
• Prototyping visual ideas for films or advertisements.
• Exploring creative video styles like photorealism or cyberpunk.

Input & Output Format

• Input: Text prompt (string), optional input image (string), optional last frame image (string), optional reference images (array of strings), duration (integer), resolution (string), aspect ratio (string), fps (integer), camera_fixed (boolean), seed (integer).
• Output: Video file (URI string).

Performance Notes

Generation speed and resource intensity are not explicitly detailed, but it is capable of producing videos in 5s or 10s durations. It can handle various resolutions (480p, 720p) and aspect ratios.