Open agent harness for creating text- and image-list-to-video generations up to 3 minutes long in a single shot. Runs in a sandboxed enterprise environment, enable only the settings and providers you need. Model providers (and models) are fully composable for each agent run. Comes bundled with Studio for optional post-processing, an API surface for custom integration and external distribution, and additional high-level generative endpoints for enterprise use cases.
Create high fidelity prompt to video with accurate prompt to video in 1-shot up-to 3 mins long. We combined with 3 video models to demo 1-shot results. Ultra high text accuracy and instruction adherence even for complex scientific topics etc.