Guide · Published August 16, 2026

How to Upload AI Designs to Printful and Printify Without Getting a Bad Print

You can't upload an AI image straight to Printful or Printify and expect a good print. The generated image is almost never print-ready: it's the wrong shape for the print area, it usually has a background baked into the pixels even when it looks transparent, and it's often smaller than the print size demands. To upload an AI design to Printful or Printify properly you need to do four things first: strip the background to real transparency, resize to the product's actual print area, save as a PNG in sRGB, then check the mockup before you publish.

That last step is the one that catches people. A file that uploads cleanly can still print wrong, and the reason has nothing to do with resolution.

What does "print-ready" actually mean to Printful and Printify?

Both suppliers publish hard requirements, and they aren't identical. Here's what each one asks for.

The number everybody quotes, 300 DPI, is a canvas size in disguise. DPI on its own means nothing until you pair it with a physical print size. A 4,500 by 5,400 pixel file is 300 DPI at 15 by 18 inches and 150 DPI at 30 by 36 inches. What both suppliers really want is enough pixels to cover the print area at the size you're printing it.

So the practical rule is simple: find the print area for the exact product you picked, in pixels, and build your file at those dimensions. Every other spec follows from that.

Why doesn't an AI image tool just give you a print-ready file?

Because it's solving a different problem. Image generators make pictures. Print files are manufacturing instructions. Four gaps show up almost every time.

Transparency is usually fake. Ask a model for a transparent background and most of them draw one. You get a gray and white checkerboard rendered into the RGB pixels, which prints as an actual gray and white checkerboard on your shirt. Even when a model does produce an alpha channel, it's often partial. The reliable approach is the opposite of what people expect: generate on a solid contrasting color, bright green works well, then key that color out afterwards. You get a clean edge and you can see exactly what you removed.

The aspect ratio is wrong. Most generators default to square. Almost no print area is square. A tee front is roughly 5:6 tall, a mug wrap is wide and short, a poster is whatever size you ordered. Dropping a square file into a tall print area means the supplier either letterboxes it, so your design prints small, or scales it to fill, so your design gets cropped.

The resolution is often too low. Plenty of models output around 1,024 pixels on the long side. That's well under a 4,500 pixel tee front. It'll upload, and it'll print soft.

Backgrounds hide in the corners. After keying, look inside enclosed shapes: the loops in letters like B, e, a and o, the gaps in a wordmark, the hole in a logo. Edge-based background removal doesn't reach them, and they print as colored blobs.

How do you upload an AI design to Printful, step by step?

1. Pick the product first, then get its print area. Printful publishes per-product templates in its design tool. Download the template for the exact item, not a generic tee, because print areas differ by garment and by placement.
2. Key the background to real transparency. Remove the solid color you generated on, then sweep the interior shapes for leftovers. Check the alpha channel rather than trusting your eyes, because a checkerboard preview in your editor can be either real transparency or a picture of a checkerboard.
3. Build the canvas at the print area's dimensions and center your artwork on it. Don't stretch artwork to fill the whole area. On a tee front, a chest print that covers roughly three quarters of the print area width, positioned with a comfortable gap below the collar, reads as retail. Filling the print area edge to edge is the most common self-inflicted design problem, and it pushes artwork up under the collar seam.
4. Save as PNG in sRGB, then upload and check the mockup at full size. Not the thumbnail. Open the mockup at full resolution and look for clipped edges, a white halo around the artwork, color showing inside letter shapes, and whether the position matches what you intended.

One extra note for dark garments: transparent pixels should carry white underneath rather than black. If your editor leaves black in the transparent regions, some printers render a dark rectangle where the transparency should be.

What changes when you upload to Printify instead?

Mechanically the process is the same, and the practical differences are worth knowing before you commit a catalog to one supplier.

Printify leans toward 300 DPI as its default recommendation and relaxes it for large-format goods. Print areas are set per blank and per print provider, and Printify routes production across many print providers, so the same blueprint can carry different print areas depending on which provider fulfills it. Check the print area in the product creator for the provider you'll actually use.

Cost visibility differs too. Printify doesn't expose per-variant base cost in the catalog data that connected platforms read, so comparing costs across Printify blanks is more manual than it is with Printful. That's a real difference, and it matters far more for your pricing than for your print files. We covered how it plays out day to day in Printful vs Printify.

Gelato behaves differently again. In our own testing, Gelato scales the submitted file to fill the print area and doesn't take separate placement instructions, which means the file effectively is the print area. Send a square design to a tall Gelato print area and it stretches or crops. Compose to the print area's exact aspect ratio and it renders correctly.

The step almost everyone gets wrong: the print area is not the visible face

This is the failure that survives every resolution check, and it's why a technically perfect file can still produce a product you can't sell.

On flat goods, a tee front, a poster, a mousepad, the print area and the visible face are the same rectangle. On anything that wraps, folds or carries hardware, they aren't. The supplier's print area describes the surface the printer covers, not what a customer sees standing in front of the product.

Real examples we've measured while calibrating garments:

• Mugs. The print area wraps the full circumference. The front-facing arc is only the middle portion of it. Center your design in the file and part of it disappears around the side of the mug.
• Tote bags. On many totes the single print area covers front and back, folded at the bottom. A centered design straddles the fold and gets cut in half.
• Drawstring bags. Same folding issue, plus a channel at the top and grommets in the corners. The visible front is roughly the upper half of the area.
• All-over-print backpacks. The front panel has a pocket seam crossing it. Lettering and faces placed across that seam get split, so artwork needs to favor the top of the panel.
• Sublimation socks. Separate front and back leg panels, and on some products the front panels render the file rotated, so one file applied to every panel prints upside down on half of them.

None of this is discoverable from a spec sheet. You find it by printing a test grid, looking at the render, and writing down where the visible face actually sits. It's also why a "just upload it" workflow feels fine on tees and quietly fails on everything else.

How ApparelHub handles the handoff

ApparelHub exists to remove this step, so here's what it does and where it stops.

You generate a design inside the platform using the leading image models we've wired in, or upload one you made elsewhere. When you pick a garment, the platform reads that product's real print area from the supplier and composes your design onto it at real dimensions, so you're not downloading templates and doing canvas math by hand. There's a transparency step for keying backgrounds, and uploads that arrive under the resolution floor get scaled up before they reach a print file. For products where the visible face and the print area diverge, the platform carries a layout for that item and places artwork on the part the customer actually sees, which covers the mug, tote, drawstring and backpack cases above instead of leaving them to you.

Then it renders the supplier's own photoreal mockup, so what you approve is what that supplier will print, and pushes the finished product out to your sales channels.

Printful, Printify and Gelato are all live today. Shopify, WooCommerce, Wix and TikTok Shop are all live as sales channels. Signup is open to everyone, and the free tier includes 25 image generations, so you can run the whole pipeline end to end before paying anything.

What it doesn't do: it doesn't choose the garment for you, and it doesn't approve the mockup for you. Those stay with you deliberately, because a misprint costs you money and no supplier refunds a design decision. ApparelHub also doesn't run an agent on your behalf. We expose the whole pipeline through an Agent API, an MCP connector and a Claude skill, so the agent you already use can drive it. That's how sellers automate the repetitive parts of this without giving up the approval step.

FAQ

What DPI should my AI design be for print on demand? Aim for enough pixels to cover the product's print area at print size, which usually works out to 300 DPI for standard apparel prints, with 150 DPI as Printful's floor on most garments. Build the file at the print area's pixel dimensions and the DPI takes care of itself.

Can I upload a JPG to Printful or Printify? Yes, both accept JPG, but JPG can't hold transparency, so your design prints with a visible rectangle of background behind it. Use PNG for anything printed on apparel.

How do I make an AI image background transparent for print? Generate on a solid contrasting color rather than asking the model for transparency, then key that color out and sweep the enclosed shapes inside letters and logos for leftovers. Verify the alpha channel afterwards, because a checkerboard preview can be either real transparency or a picture of one.

Why does my design look right in the file but wrong on the mockup? Usually because the print area and the visible face aren't the same surface, which happens on mugs, totes, bags and anything that wraps or folds. Check the full-size mockup rather than the thumbnail, and if artwork is clipped or split, reposition it inside the visible portion of the print area.

Do I need a different file for each supplier? Often yes, because print areas differ by supplier, by blank, and sometimes by print provider within the same supplier. If you sell one design across multiple print providers, plan on composing per product rather than reusing a single master file.

The short version

The gap between an AI image and a sellable product is small, repetitive and unforgiving: key the background, match the print area, hold the resolution, then verify against a real mockup. Do it by hand once and it's manageable. Do it across forty designs on three suppliers and it becomes the reason most sellers stall.

If you'd rather not redo that math per product, start free and let the platform compose the print file against the real print area, or point your own agent at our agent surface and have it run the whole loop.

Sources: Printful print file requirements, Printful DTG file guide, Printify print file requirements.