Agnes Video Generator combines self-hosted tools with cloud video creation
Agnes Video Generator is an MIT-licensed application that creates multi-scene videos from text through a web interface. It supports image-to-video generation, keyframe animation, digital anchors and manuscript-based workflows, alongside built-in narration and automatic subtitles. The software can run on an ordinary computer, while AI processing takes place in the cloud through Agnes AI.
The project offers generation without subscriptions or usage charges, although requests are subject to a 16 req/min rate limit. Users need a free Agnes AI API key and can deploy the application manually, through Docker or through npm. A browser-based option provides a simpler experience without installation. Local workflows also support checkpoint recovery and regenerating selected scenes.
A separate agent skill can install the service, generate videos, deliver files and diagnose failures through coding assistants. With user approval, it can also help fix defects and open pull requests. The underlying video model still has limitations: generated frames can be unstable, and complex actions can produce deformations.