What it does
This page covers Seedance 2.0, a video generation model from ByteDance, offered through fal, which describes itself as the official Seedance 2.0 API. The model takes text, image, audio and video inputs and returns MP4 video with synchronized audio, including multi-shot sequences, from a single generation. Endpoints cover text-to-video, image-to-video and reference-to-video, each with a standard and a fast version. The fast tier offers lower latency and cost, while the standard tier prioritizes quality for final renders. Reference-to-video accepts up to 9 images, 3 videos and 3 audio clips as inputs. The FAQ gives 480p and 720p output, clips of 4 to 15 seconds, aspect ratios from 21:9 to 9:16, and generation times under two minutes. Accepted inputs are JPEG, PNG and WebP images, MP4 and MOV video, and WAV and MP3 audio. Developers create an account, get an API key and call the REST API or use the Python or JavaScript client. The site also mentions a playground that needs no setup or key. US-hosted inference is offered to US companies at 480p and 720p with separate pricing arranged through fal's team, and the page notes that a newer Seedance 2.5 is also available on fal.
Capabilities
- Text-to-video, image-to-video and reference-to-video endpoints
- Audio generated with the video, including dialogue with lip-sync
- Multi-shot clips of 4 to 15 seconds with prompt-driven camera moves
- Standard and fast tiers trading quality against cost and latency
- Python and JavaScript clients plus a REST API
- US-hosted option at 480p and 720p
When to use it
- You need short AI video clips with generated sound from a script or shot list
- You want to animate still images or build on reference images, video and audio
- You are adding video generation to an app and want an API with SDKs
- You want the fast tier for lower latency and cost in production
