Reference inputs for AI video can now include web pages and documents following a new model deployment on the fal platform. Wan 3.0, developed by Alibaba, produces native 30 second clips in a single pass, removing the need to stitch multiple shots together to reach that length. The tool also supports text, image, audio, and video as reference materials for greater creative control.

The release includes reality grade rendering and improved real world motion for higher fidelity outputs. This updated version is now live to provide continuous video without the continuity errors often associated with composite shot generation.

Sign in to suggest edits

Key sources

  1. SOURCE@fal“Omni-reference input: text, image, audio, video, web pages and documents”x.com
  2. SUPPORT@sahilvermaai“No stitching two shots together and hoping the face survives the cut”x.com
  3. SUPPORT@odinlovis“really an incredible model in term of quality, respecting so well your input”x.com
Markdown