Newsletter
Join the Community
Subscribe to our newsletter for the latest news and updates
JustOG turns any AI image into a real, made-to-order garment. Bring art from ChatGPT or Gemini, or make it here, and the printed shirt, tote, poster or mug arrives in the post. The mockup you approve is what ships.
Submit your own product to reach creators and founders looking for the next tool to try.
JustOG turns any AI image into a real, made-to-order product. You can describe the design in plain language and let the studio draw it, or bring an image you already made somewhere else — ChatGPT, Gemini, Midjourney, a scan of something you drew. Either way you end up looking at the design on the actual garment, and if you like it, that garment gets printed and posted to you.
The catalogue is shirts, totes, framed prints, posters, caps, mugs and clogs. Everything is made to order, so a run of one is completely normal. There is no minimum, no inventory sitting somewhere, and nothing gets printed until someone actually wants it.
The preview is the product. On most print-on-demand sites the image you approve is a picture of a shirt — often itself AI-generated — and the thing that arrives is a different object that happens to share a colour. We do it the other way round: the design is composited onto photographs of the real blank, warped by a displacement map built from that garment's own fabric, and placed inside a print area we measured for that specific blank and that specific size. The same numbers drive the preview and the print job, because they live in one file. If the mockup shows the print sitting two centimetres below the collar, that is where it lands.
That sounds like a detail. It is the whole difference between a preview and a promise, and it is why we spent more time on garment calibration than on the generation pipeline.
Next.js 15 on Vercel for the storefront and the studio, Supabase for auth and data, Stripe for checkout. Behind that sits a small Node and Python service that does the actual image work — cutout, alpha trimming, compose, displacement — with NumPy, Pillow, SciPy and OpenCV. Image generation runs through gpt-image-2 and gemini-2.5-flash-image, with a prompt-rewriting layer in front that turns a casual sentence into something a print model can actually execute.
People who want one specific thing that no shop sells: an inside joke, a memorial, a reunion, a band that only three people remember, a world that never got merchandise. Creators who want their own art on something physical without setting up a store. And founders, who get their own door at /founder — put in a company domain and get back shirts made for that company.
Free to design and preview. You only pay when you want the real thing.
