AI product photography from the photo you already have

Shoot the reference before you open the generator

The model edits from your photo, so your photo sets the ceiling. A blurry label stays blurry in a marble bathroom. A cap that was cropped out of frame comes back as whatever cap the model invents. Ten minutes with a phone by a window does more for the final images than any prompt.

Put the product on a plain surface, switch off the ceiling lights and shoot with daylight coming from one side. Fill most of the frame but leave a little space on every edge. If the back or the base matters (an ingredients panel, a hidden zip), photograph it separately. The model can’t know what it has never seen, and it will guess with total confidence.

Studio, lifestyle or close-up first

Decide by where the image is going. A marketplace main image needs the studio shot on white, and nothing else will do. A product page needs one studio shot for the first impression and then images that answer questions: how big is it, what does the surface feel like, where does it live. Social posts and ads get skipped when they look like catalogue photos, so lead with a scene there.

If you are doing one product on a small budget, make the studio shot first, then one lifestyle scene, then one close-up. That covers a listing, a product page and a first ad. Holiday scenes and flat lays can wait until you know the product sells.

A leather card wallet, start to finish

Say you sell a slim card wallet in tan leather, and your only photo shows it on a desk with a mug in the corner. Re-shoot it on a sheet of white paper by a window, front and back, plus a third close-up of the stitching along one edge. That is your reference set.

For the studio image, ask for the wallet alone on white, filling most of the frame. For the lifestyle image, describe a scene rather than a mood: “on a walnut desk beside a closed notebook, morning light from the left, shallow focus.” For the close-up, name the detail: “macro of the edge stitching and the stamped logo.”

Then hold each result next to the real wallet in daylight. Tan leather drifts toward orange or brown easily, and a shade off is a return waiting to happen. If the stitch spacing or the logo changed, regenerate from the close-up reference rather than the full front view; the model can only keep what it was shown clearly.

Making twenty products look like one store

What makes a store look put together is usually consistency, not polish. Once a look works, write it down as a sentence you can reuse: the surface, the light direction, the background colour and how much of the frame the product fills. Use that sentence for every product in the collection and change only the product.

Keep props to one or two and keep them generic. A sprig of eucalyptus suits candles, serums and soap; a branded mug in the corner suits nothing. Keep the light coming from the same side across the set, too. Shoppers notice when shadows fall left on one product and right on the next, even if they couldn’t say what feels off.

Shoot the references the same way, too: the same window, the same time of day, the same sheet of white paper. Consistent outputs are much easier to get from consistent inputs, and a reference set you can repeat is what lets you add product twenty-one next spring without the collection looking patched together.

Where AI photos are the wrong choice

Colour-critical products need real photos under controlled light: paint, fabric sold by the metre, foundation shades. A generated image can look right on your screen and still be a shade out. Clothing fit is similar. A generated scene can put a jacket on a model, but it can’t show how your size M sits on a real body.

Glass, chrome and other clear or mirrored surfaces are the hardest to keep faithful, because the model redraws the reflections. For those, let AI handle the background and setting, and keep a real photo of the product wherever a buyer will decide on the exact detail. Platforms expect that anyway; TikTok Shop’s product rules, for one, say images must accurately represent the product the customer will receive.