Unlike standard text-to-video tools, the O3 model allows for Video-to-Video editing, where you can modify existing footage by restyling the scene, swapping props, or relighting shots while maintaining the original motion and character performance.
Key Features
- Video-to-Video Transformation: Reshape existing footage (e.g., transforming a daytime street into a neon-lit cyberpunk city) while keeping motion intact.
- Subject & Prop Swapping: Replace specific objects or character features (like a "Prop Swap") using image references.
- Relight: Use specialized VFX controls to change the lighting direction of a generated scene.
Technical Capabilities
- Modalities: V2V, with optional Elements and Reference Image
- Ratios: 1:1, 16:9, 9:16
- Durations: Supports 3 to 15 seconds
- Language Support: Native support for English, Chinese, Japanese, Korean, and Spanish.