howseen-ai/claude-motion-design

A Claude Code skill to make motion design videos in pure code: HTML + Playwright + ffmpeg. No After Effects, no Remotion license.

What it solves

It enables the creation of high-quality motion design videos using pure code (HTML, Playwright, and ffmpeg) instead of traditional video editing software or expensive licensed tools. This approach makes video production deterministic, free, and easily iterable by AI agents, allowing for precise beat-syncing and rapid revisions.

How it works

The system uses a seek(t) function where every frame is computed based on time, ensuring consistency. It renders frames via Playwright screenshots and compiles them using ffmpeg. The workflow includes a beat map for music synchronization, subframe blending for motion blur, and a two-pass audio normalization process. It also includes a "Remake mode" that can analyze an existing launch video to extract shot specs and recreate a branded version using AI agents.

Who it’s for

Developers and AI users (specifically those using Claude Code) who want to generate professional-looking promotional or launch videos without needing expertise in After Effects or Remotion.

Highlights

  • Deterministic Rendering: The entire film is treated as a function of time, allowing specific scenes to be fixed without affecting the rest of the video.
  • Beat-Synced Audio: Music drops and sound effects are aligned to measured peaks and band energy rather than generic beat grids.
  • AI-Native Workflow: Designed as a Claude Code skill, allowing users to request changes like "make transitions faster" through simple code edits.
  • Remake Mode: A specialized pipeline that extracts frames from a reference video to create a 1:1 branded remake.
  • Zero-Cost Tooling: Relies entirely on open-source tools like Chromium, ffmpeg, and Python.

Related

  • Project
    SamurAIGPT/AI-Faceless-Video-Generator

    An automated pipeline that generates talking-head videos from a topic by combining GPT script generation, text-to-speech, and SadTalker face animation.

  • Project
    NVIDIA/flashdreams

    FlashDreams is an Apache‑2.0 library for ultra‑fast inference of autoregressive video and world‑model networks, targeting real‑time applications like gaming, autonomous‑vehicle simulation, and robotics. It provides ready‑made runners for several NVIDIA models, supports multi‑GPU setups on H100‑class hardware, and offers extensible guides for adding new methods.

  • Project
    Anil-matcha/Open-AI-Micro-Drama-Generator

    An agentic AI pipeline that transforms a simple idea or script into a complete cinematic micro-drama video by automating story development, character design, and video production.

  • Project
    OpenGVLab/VideoChat-Flash

    VideoChat-Flash is a multimodal large language model designed for efficient long-context video understanding, capable of processing videos up to three hours long using hierarchical compression.