How to put a ChatGPT image on a T-shirt

You've been in the same thread for forty minutes. The first six images were wrong in ways you couldn't name until you saw them. The seventh is right — the composition, the color, that one detail you didn't know you wanted.

Then it sits there. A square in a chat log.

Search for what to do next and you'll find plenty of advice: check your DPI, upscale it, remove the background, convert the color profile. Most of that describes something real. Very little of it is actually your job. This is what happens between the file and the shirt, and which part is whose.

The 60-second version

What only you can do:

• Hand over the original download, not a screenshot. Nothing downstream can undo this one.
• Read the words, if there are words. AI-generated type is confidently, subtly misspelled, and only you know what it was supposed to say.

That's the whole list.

What JustOG handles:

• The prompt, if you're making one here. Describe the idea in a sentence; it gets expanded into several fully-specified directions, with a note on how it read you.
• Resolution. Sizing your file for the press, upscaling included.
• The background. Cutting the artwork off whatever square it arrived on.
• The framing. Where the art sits and how big it reads. You drag the frame; nothing gets regenerated.
• How it's made. Which process puts your design onto the garment, and how it behaves there.

Can you print an image ChatGPT made?

Yes. OpenAI's terms of use assign you the rights to the images you generate, and people sell AI-designed apparel commercially every day.

That covers OpenAI's claim on the file, though — not everyone else's. If you prompted your way to a recognizable character, a studio's poster, a band's logo, or a brand's mascot, that's still someone else's intellectual property, and who generated the pixels doesn't change it. Your own photo, your own dog, your own words, an original composition in a style you like: all fine. A named franchise: not fine, whatever the model was willing to draw. (Ours is the same rule — see the content policy.)

Why do people say AI images print blurry?

Because of a number that rarely comes up outside a print shop: pixels per printed inch.

An image model hands back something around 1,024 pixels on a side. A big chest print is on the order of a foot across. Divide, and that file is landing at something like 85 pixels per inch across the print, where print work generally starts getting comfortable around 150 and looks crisp at 300. On a screen you'd never know.

The gap is real, and it's ours to close, not yours: your file gets sized and enlarged for production before anything touches a garment. Redrawing is part of that toolkit too — an image regenerated at print size isn't your file made bigger, it's a new one that takes its likeness from yours. (More on that in turning a photo into a shirt.)

What you do have to hand over is the real file, which is the next section.

Should I screenshot the image or download it?

Download it. Use the download button and keep the original.

This is the one place where the damage can't be repaired later. A screenshot is captured at your display's size rather than the image's — often half the pixels — then re-compressed, sometimes with a sliver of interface or a rounded corner baked into the frame. Upscaling enlarges what's in the file. It can't bring back what the screenshot discarded, and if a menu bar made it into the frame, that gets sharpened too.

While you're there, save the prompt. If you ever want the design in another colorway, on a different garment, or as a matching back print, the prompt is worth more than the image.

Do I need to remove the background?

No. Separating your artwork from whatever it was sitting on is part of the job here — you upload the square, and what goes on the shirt is the art.

If you want to make that separation cleaner, there's one thing that helps, and it's the opposite of what most people try. Ask for a flat, evenly lit solid background rather than a transparent one. Image models are unreliable at transparency: you'll get a painted-on checkerboard, a gray haze where the alpha should be, or a soft halo of background color clinging to every edge. A plain solid backdrop in a color that appears nowhere in the artwork separates far more cleanly. Something like: "the subject centered on a flat, even, solid background, no shadow, no gradient, nothing else in frame."

Check the words before you print them

Current models spell. The advice that image generators can't write text is a couple of generations out of date — short slogans, distressed and chrome lettering, even long passages come back correct now.

The thing that hasn't changed is mechanical: re-rolling an image to fix one word redraws all of the type, so the letters that were fine drift while you chase the one that wasn't. Fix the word in the prompt, not by rolling again and hoping.

And read it once before it goes on a shirt. Not because the spelling is likely wrong — because nothing automatic knows what you meant to say, and a correctly-spelled wrong word is the one failure that survives every check but yours.

What changes on a dark shirt?

Mostly the look, which is the part you get to choose.

A design on black reads differently from the same design on bone — deep tones in the art sink into a dark garment and the light areas do the work, which is often the better version and occasionally the wrong one. That's a taste call, and the preview makes it before you buy: your design, the actual garment, the actual color.

What doesn't change is what's possible. There isn't one printing process with one set of limitations — ink, foil, dye and thread all put an image onto cloth differently, and picking the one that serves a particular design on a particular garment is our end of the job, not a constraint you design around.

Where does the design sit on the shirt?

A big chest print runs most of the width of the front and starts a few inches below the collar. A square design fitted to that width lands as a centered statement; a tall one scaled to fit the same height reads much smaller.

You don't go back to the model for that. You drag the crop frame over your image until the art sits the way you want, and the preview shows it composited on the actual garment at the actual size before you order.

A big chest statement is one option among several. Placement is a design decision in its own right: a dense, detailed scene needs the whole front, a single bold mark reads better small and high, a tall composition belongs on a back. Blanks differ the same way — a sleeveless cut, a boxy one and a cropped one present three different fields — so part of the work is picking the garment and the placement that suit the image.

How do I actually get one made?

Bring the file. On JustOG you upload the image you made in ChatGPT — or Gemini, Midjourney, anywhere — see it on the real garment, pick the color and the size, and the piece is printed to order and shipped to you. One is a normal order; there's no minimum and nothing is held as stock.

If you'd rather start from something that has already been through all of the above, the shop is designs other people made and published.

That image has been sitting in a chat log long enough.