Set up your brand or channel once. Content Factory finds what your audience wants to watch, researches it, writes the reel in the form your channel uses, voices it and animates every frame, on your own machine and your own Claude plan. A reel that does not clear the bar is never posted: it waits for you, with the reason.
Picks the topic Live trends, Reddit, Hacker News, news and Wikipedia, read for your audience and voted on by a panel of strangers before a word is written.
Checks every fact Each number and quote is matched to the source it came from before the voice is recorded.
Draws every frame No image or video generators. Two film engines and 90+ looks, plus real, credited material from the open web.
Judged before it posts Scored against hand-picked reels. Under 8 out of 10, it is fixed once, then held for you.
The studio: the reel being made, what waits for you, and when the next one starts.
Every frame is made, not generated
Cel a vault door
Linocut why snow keeps houses warm
Waterfall where a salary goes
Machine four checks before the origin
Cyanotype how plants travel
Map where one tap goes
Mechanism same request, same key
Cel RSA-2048, struck
Riso interest grows on interest
Cutaway how a winch holds its load
Motion −30 dB of engine hum, gone
Sketch why it rises
A reel moves through several of these, each beat in its own look, joined by designed transitions.
Watch it being made
Watch any reel being made, step by step, and open every brief, script and still behind it.
Seven steps from a question people already have to a reel you can post: the facts, the hook, the script, the pictures, the sound and a fresh-eyed check, with prompts to paste. Free, no sign-up, made from what we learned building Content Factory.
Open source
Tools we built while making Content Factory. Free under the MIT licence, no dependencies, one npx away.
Finds the flashes that can trigger photosensitive seizures (WCAG 2.3.1), captions nobody can read and sudden loudness jumps (EBU R128) in any video, before you post it.
One search across Wikimedia Commons, NASA, the Smithsonian, three museums and Poly Haven, keeping only what each item's own licence lets you reuse, with the credit line written. CLI and MCP server.
Word-by-word captions, the kind short-form video uses (karaoke, highlight, reveal, pop), from a Whisper or other transcript, burned in by ffmpeg in one command.
There was an error while loading. Please reload this page.
Most used topics
Loading…
Uh oh!
There was an error while loading. Please reload this page.
You are visitor 01402503 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required.