gatewai-dev/framefields

Code-first video & image, rendered natively on WebGPU.

What it solves

Framefields provides a high-performance, code-first way to generate video, motion graphics, and 3D scenes. It is specifically designed to be controlled by AI coding agents, allowing them to programmatically create complex visual assets without the overhead of traditional headless browser-based rendering.

How it works

Instead of using a browser engine (like Puppeteer or Chromium) to render frames, Framefields uses native GPU hardware via WebGPU, Dawn, Metal, and Vulkan. This allows for deterministic, frame-accurate rendering of TypeScript-defined compositions. It integrates directly into development environments like Cursor or Claude Code, enabling agents to write code, inspect frame outputs, and export high-quality MP4 files.

Who it’s for

This tool is for developers and AI agents tasked with automating video production, motion design, or 3D rendering workflows. It is ideal for those who need precise control over visual output and want to avoid the performance bottlenecks and non-deterministic behavior of browser-based automation.

Highlights

  • Native GPU Rendering: Bypasses headless browsers for faster, more reliable rendering using WebGPU and native graphics APIs.
  • Agent-Optimized: Designed to be installed as a skill for AI coding assistants like Claude Code, Cursor, and Copilot.
  • Unified Creative Suite: Combines compositing, motion graphics, typography, and 3D scene management into a single TypeScript package.
  • Deterministic Output: Ensures frame-accurate timing and audio synchronization, critical for professional-grade video production.
  • Built-in Vision Capabilities: Includes support for on-device ONNX models for tasks like object tracking and pose estimation.

Related

  • Project
    SirLynix/obs-kinect

    obs-kinect is an OBS Studio plugin that lets you use any Kinect camera as a video source and create a virtual green screen effect based on depth or body detection, removing your background without a physical green screen. It supports Kinect v1, v2, and v3, and runs the heavy processing on the GPU for good performance.

  • Project
    trustfuture/simon-skills
  • Project
    jnMetaCode/ai-shortfilm-prompts

    A genuine open‑source project that provides a library of AI video‑generation prompts, a 5‑stage prompt methodology, a web‑based prompt builder, and a Claude Code skill for turning ideas into ready‑to‑run multi‑shot prompts for models like Sora, Seedance, Veo, Kling, etc.

  • Project
    yizhi-chengzi/video-ai-talking

    A local web app that turns a silent face‑cam clip into a full talking‑head video using Alibaba Bailian VideoRetalk for lip‑sync, TTS from Bailian or Volcengine, optional DeepSeek script generation, and FFmpeg for final rendering.