PDF · Free download
Keep this guide as a PDF
brief2mp4: The Full Guide to Making Videos from a Written Brief, as one file you can read offline and share with your team.
Sign in with Google or GitHub to download.
This is the companion guide to the video Fable 5.2 vs Opus 5.5 vs GPT-6.1: Same Brief. Three AI models got the same motion-graphics briefs in four rounds. At the end of the video I show the free skill that turns a written brief into a finished, checked MP4. This guide has every prompt and every setting.
Install it. One line in a terminal:
git clone https://github.com/qualixar/brief2mp4.git && bash brief2mp4/install.sh
No terminal? Open Claude Code and type: Install the brief2mp4 skill from github.com/qualixar/brief2mp4 for Claude Code and Codex. Then start a new session, because a session only sees skills that existed when it started.
Make a video. Type Use the brief2mp4 skill to make the video in this brief: and paste your brief. When it says it is done, send one follow-up: Check it against the brief. The skill re-reads your brief line by line, shows the evidence for each line, fixes what fails and renders again.
You need Claude Code or Codex on your own Claude or ChatGPT plan, Node.js 22 or newer and ffmpeg (Blender only for Blender briefs). The skill is free and MIT licensed: github.com/qualixar/brief2mp4.
What is inside the 69-page PDF:
- The whole system: the loop (spec, route, compose, finish, check, deliver) as diagrams, and why time is the only animation input.
- Setup on macOS, Windows and Linux: one message an agent follows, the doctor, and your first proof render.
- Writing a checkable brief: the anatomy of SPEC.md and five annotated example briefs with their real measurements.
- The format atlas: beat sheets and routes for 2D explainers, maths, tutorials, stories and skits, Shorts, data stories, product launches, logo stings, kinetic type and captions.
- The 3D pack and the ArtCraft edit pack: components, shot JSON, baked physics, verified downloads and editing recipes.
- Self-check, sound and finishing: delivery checks, contact sheets, phone safe zones and the loudness math.
- The ecosystem directory: the open-source repos worth knowing, with licences and when to use each.
- From the video: how the contest was run, stills of what each model made, the toolbox and all four contest briefs word for word, and the results and receipts.
- Troubleshooting, FAQ, glossary and a one-page cheat sheet.
Watch the film: Fable 5.2 vs Opus 5.5 vs GPT-6.1: Who Makes Better Motion Graphics? · Read the deep dive: I Gave Three Frontier AIs the Same Video Brief
One run per model; results can vary run to run. Read the output back before you publish it. That habit is what we call AI Reliability Engineering.