video▌

通过视觉/视频大模型分析视频内容，支持本地视频文件和互联网视频。

Generate talking head and avatar videos from images and audio using OmniHuman, Fabric, and PixVerse models. \n \n Four model options: OmniHuman 1.5 (multi-character), OmniHuman 1.0 (single character), Fabric 1.0 (image lipsync), and PixVerse Lipsync (highly realistic) \n Audio-driven workflow: pair portrait images with speech files to generate realistic avatar videos with synchronized lip movement \n Composable with text-to-speech and video transcription for end-to-end pipelines: generate speech

Create AI avatar videos with full control over avatars, voices, scripts, and backgrounds using POST /v3/videos. Two creation modes via discriminated union on type:

Convert React/Remotion component code directly to MP4 videos via CLI. \n \n Accepts TSX code with full Remotion API support: useCurrentFrame , useVideoConfig , spring , interpolate , AbsoluteFill , Sequence , and media components \n Configurable output: resolution, FPS, duration, codec, and component props passed as JSON \n Supports streaming progress updates and Python SDK integration for programmatic video generation \n Ideal for animated graphics, motion design, data-driven videos, and React

Generate videos from text or images using Pruna's optimized models via inference.sh CLI. \n \n Three models available: P-Video (text-to-video, image-to-video, audio support), WAN-T2V (fast text-to-video), and WAN-I2V (image animation) \n Supports 480p, 720p, and 1080p resolutions with draft mode for faster, cheaper generation \n Audio synchronization available for P-Video to create videos with synced audio tracks \n Requires inference.sh CLI ( infsh ) and authentication via infsh login \n

Keyframe, skeletal, and morph target animation system for Three.js objects and models. \n \n Supports four animation track types: number, vector, quaternion, and color keyframes with linear, smooth, and discrete interpolation modes \n AnimationMixer plays clips on objects with full playback control: play, stop, pause, speed adjustment, weight-based blending, and fade in/out transitions \n Skeletal animation via bone manipulation, bone attachments, and GLTF model loading with automatic clip disco

Assemble course videos from individual lesson files with title slides and auto-generated timestamps for YouTube.

Generate videos with 40+ AI models including Veo, Seedance, Wan, and Grok via inference.sh CLI. \n \n Supports text-to-video, image-to-video, avatar animation, lipsync, video upscaling, and foley sound generation across multiple model families \n Access 10+ text-to-video models (Veo 3.1, Seedance 1.5 Pro, Wan, Grok Video) and 5+ image-to-video variants optimized for speed, quality, or cost \n Includes avatar and lipsync tools (OmniHuman, Fabric, PixVerse) for talking-head and character animation