Features
Declarative JSX video composition (no React dependency)
Unified API for video, image, voice, music, and lipsync generation
Content-addressed caching (instant reuse on identical props)
16 core primitives: Render, Clip, Image, Video, Animate, Speech, TalkingHead, Music, Title, Subtitle, Captions,
Automatic FFmpeg rendering server-side
Error messages designed for AI agent debug (actionable hints)
TypeScript first with type-safe props
CLI tool (varg render) to output MP4 directly
Supports multiple AI providers: fal.ai, ElevenLabs, OpenAI, Replicate, Higgsfield, Kling, Flux
Templates for common patterns (slideshow, talking character, transformation)
Native 4K video generation (no upscaling)
Camera controls: push, pull, pan, tilt
Text-to-video up to 15 seconds
Image-to-video generation
Native Audio in Kling 3.0 Omni
Multimodal instruction parsing
Prompt weighting for fine control
Director style prompts (e.g., Wes Anderson)
Series mode for multi-image consistency
Subject binding across shots
API access for developers
Desktop app (macOS/Windows)