Share your video's script and lean back — bi{VARYASYON} handles the rest.
Paste your timestamped transcript and hit "Split into lines". Every line becomes a variation, every timestamp becomes a filename.
Place the {VARYASYON} tag inside your base prompt, then check the outgoing prompts in the preview.
Log in to ChatGPT in the browser Playwright opens, then start. Images are generated one by one and you follow progress in the run log.
Filenames carry your edit timeline: show each image from the previous cut-off up to the time in its own name, and it stays in sync with the audio.
The live run log shows the status of every line; pause, resume or stop.
Playwright opens a browser on your computer; generation happens there, with your own ChatGPT account. Your session and your limits stay yours.
All the code is on GitHub. Read it, audit it, adapt it to your needs, or contribute.
No subscription, no API key, no hidden fees. Your existing ChatGPT account is enough.
Everything runs on your own machine. Your scripts, prompts and images never leave your computer; no external server.
Copy the commands into your terminal, or hand the prompt to an AI agent.
Free and open source.
Paste your script, hit Start — your visuals are done.