Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseResearch Paper · For R&DX: @flxrnc→ original

15 Free Repositories Teach an AI Agent to Copy Animation: 99 Styles and 244 Effects

Want an AI agent to recreate an animation you like without paying for a professional video editor? An X author compiled 15 free repositories for this. Remotion builds video in code, the way you would build a website. Blender MCP lets the agent control the Blender 3D editor. The collections include 99 film styles, 244 pixel effects, and 703 ready-made prompts. A practical starting point is Remotion with Remotion Skills, tested on one short scene.

AI-processed from X: @flxrnc; edited by Hamidun News
15 Free Repositories Teach an AI Agent to Copy Animation: 99 Styles and 244 Effects
Source: X: @flxrnc. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceX: @flxrnc ↗
Editor-in-Chief
Lemo-Opuscar99 film styles, all "shot" in code
PixelFX244 ready-made retro effects in pixel style
Key metric #3The author shows no quality benchmarks; check the licenses yourself

Imagine this: you see an animated video on social media and want to make something similar for your product, but you don't want to pay for a professional video editor or manually move layers on a timeline. The idea behind a list posted on X by the author @flxrnc is this: if an AI agent can write code, it can "shoot" video the same way it builds websites. To do this, the author compiled 15 free repositories.

They include tools that analyze an existing video and rebuild it, collections of 99 film styles, 244 pixel effects, and 703 prompts, plus libraries for 3D, text animation, and design polishing. The agent receives the entire list and is asked to recreate a favorite video. Below, we look at what's on the list and what can be trusted.

An important caveat. The list promotes the author's longer analysis of the Fable 5.2 model and "9 lessons from 40 ready-made films," but the article itself is cut off in the source. We therefore cannot judge what Fable 5.2 is capable of. The post contains no data on output quality. It is a catalog of tools, not a test.

What analyzes video and sets the style

  • Rea-Video breaks down any video into its components and recreates it. The entire list is built around this "copy of an animation" use case.
  • Lemo-Opuscar contains 99 film styles, all "shot in code": the style is defined by text and parameters, not by manual editing.
  • EffectCraft offers motion graphics and visual effects without paid software.
  • PixelFX provides 244 retro effects in pixel style that can be inserted into a video as ready-made elements.
  • Awesome Opus 5.5 Videos is a collection of 703 prompts taken from popular AI videos. It is useful as a bank of examples for showing the agent what a good scene description looks like.

What turns code into images

  • HyperFrames from HeyGen plays back HTML written by the agent as video. For a model that writes web pages well, this is the shortest path to a video.
  • Remotion builds video with the same code used for websites (in React). The video becomes a project that can be stored in git and edited in a regular code editor.
  • Remotion Skills are official guidance for working with Remotion, which the agent, according to the author, actually follows. It is a set of instructions that keeps the model from inventing the API from scratch.
  • Blender MCP connects the Blender 3D editor to the agent through the MCP protocol (a standard way to give a model access to external programs). Blender executes what the model tells it to do.
  • Three.js is a well-known library for real 3D in the browser on any device.
  • GSAP handles text animation: letters fly in, bounce, and glide.
  • ComfyUI generates AI images and video with full control over the process through nodes.

The last three items are there so the result doesn't look "machine-made." Unslop turns templated AI design into something more lively, Impeccable brings the styling up to a professional level, and Anthropic Skills are Anthropic's own Claude skills for design and code.

What to do with this in practice

In theory, the chain works like this. Rea-Video analyzes the source, Remotion or HyperFrames assemble it in code, GSAP and Three.js add motion and volume, and the style and prompt collections define the appearance.

For developers and product managers, the list is useful as a map of the ecosystem. Remotion, Three.js, GSAP, ComfyUI, and Blender MCP have existed as separate projects for a long time. What's new here is the idea of connecting them through an agent and skills: text instructions loaded into the model for a specific task.

There are limitations, too. The author shows no benchmarks or comparisons with original videos, so the claim that it "can copy any animation" remains promotional until there are examples. Licenses for third-party repositories need to be checked. Copying another creator's video style for style may raise copyright concerns. Free code doesn't eliminate the costs of the model, GPUs, and debugging time.

A reasonable start: take Remotion together with Remotion Skills, give the agent one short scene, and see how many revisions it takes to reach an acceptable result. If that works, add Three.js or Blender MCP. Connect the other items on the list only when a specific task calls for them, not all at once.

Story

Key Questions Answered

What analyzes video and sets the style?

Rea-Video breaks down any video into its components and recreates it. The entire list is built around this "copy of an animation" use case. Lemo-Opuscar contains 99 film styles, all "shot in code": the style is defined by text and parameters, not by manual editing. EffectCraft offers motion graphics and visual effects without paid software. PixelFX provides 244 retro effects in pixel style that can be inserted into a video as ready-made elements.

What turns code into images?

HyperFrames from HeyGen plays back HTML written by the agent as video. For a model that writes web pages well, this is the shortest path to a video. Remotion builds video with the same code used for websites (in React). The video becomes a project that can be stored in git and edited in a regular code editor. Remotion Skills are official guidance for working with Remotion, which the agent, according to the author, actually follows.

What to do with this in practice?

In theory, the chain works like this. Rea-Video analyzes the source, Remotion or HyperFrames assemble it in code, GSAP and Three.js add motion and volume, and the style and prompt collections define the appearance. For developers and product managers, the list is useful as a map of the ecosystem. Remotion, Three.js, GSAP, ComfyUI, and Blender MCP have existed as separate projects for a long time.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Hands-On AI Engineering · Hamidun Academy

Build & Ship Production AI Agents Without Boilerplate Code

Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.

Start 72h Free Trial →Vibe-Coding Intensive
What do you think?
Loading comments…