Competitor photos are the most underrated raw material in image work. Not to copy — to borrow the light. A competitor has already paid for the shoot, solved the reflection problems, and figured out which angle makes the product look expensive. What they haven't done is shoot your product. That's the gap this technique fills.
The method is a style transfer pass done with restraint, using an image-to-image workflow like the one on the image to image homepage. You feed in your own photo and a reference, and you pull the lighting and color character from the reference while keeping your product's shape and detail. Done well, nobody can tell it wasn't shot on the same table.
Step one: pick references for light, not for product
This is the part people get wrong. The best reference is usually a competitor's photo of a different product shot in the same conditions you want — same background family, same light direction, same level of contrast. If you grab a reference of a similar product, the transfer starts pulling surface qualities across and your product stops looking like your product.
Build a small folder of five to ten references per category. Sort them by lighting: hard directional, soft window, overhead flat, backlit rim. When you need a new listing, pick the lighting you want, not the object you envy.
Step two: match your shot to the reference first
Before any AI runs, get the two images roughly aligned in composition. If your product is shot from above and the reference is at eye level, the transfer has to invent geometry, and it will invent badly. Reshoot or re-crop so the product occupies a similar portion of the frame and the horizon sits at a similar height.
Do this with a real camera when you can. A phone on a tripod with a lamp and a piece of white foam board gets you 80% of the way to a studio setup, and the AI only has to close the last 20%. The less correction the transfer has to do, the fewer artifacts it leaves behind.
Step three: turn the strength down lower than feels right
The instinct is to push strength high so the result looks "finished." Resist it. Every point you add pulls more of your product's actual pixels away and replaces them with the model's guess. On product work, the sweet spot tends to sit surprisingly low — enough to shift the color temperature and the highlight behavior, not enough to redesign the label.
Run a ladder: three outputs at three different strengths, same seed, same reference. Put them next to the original and zoom to 100%. Look at the edges of your product, the text on the packaging, and any fine texture. The moment those start going soft or getting "improved," you've gone past the useful point.
Step four: protect what has to stay exact
Anything a customer relies on to identify the product should be checked at full size after every pass — the label, the logo, the shade of the product itself, the number of buttons, the stitching pattern. If the transfer changed any of it, mask that region out of the pass and composite the original back in afterward.
This is boring, mechanical work and it's the whole job. The transfer is the easy part. The verification is what separates a listing photo that survives a return request from one that creates a dispute.
Step five: check color against a physical reference
Screens lie, and so do models. If the product color matters for the sale — and for anything sold as a specific shade, it does — put a physical color swatch in the frame during the original shoot, or compare the finished image against the product under neutral light. The transfer will happily shift a dusty rose into a peach because the reference had warm light.
Fix color drift in the source photo's white balance before running the transfer, not after. Correcting it afterward means re-running the whole pass, and the second run won't match the first.
When this technique is the wrong call
Three situations where you should shoot for real instead. First, anything with regulatory labeling where the image must be exact. Second, highly reflective or transparent products — the transfer will produce a beautiful reflection that isn't physically possible, and buyers who receive the item will notice. Third, when your competitor's lighting is actually bad. Copying a trend that isn't working for them doesn't help you.
Also worth saying plainly: this is a lighting and color technique, not a shortcut for product quality. A poorly made item photographed with borrowed studio light still looks like a poorly made item, just a better-lit one.
A workflow you can repeat
Keep a folder per category with your approved references. Keep a text file with the strength setting and seed that worked. Shoot your product as cleanly as you can, match the composition, run the ladder, verify at 100%, and composite back anything that got damaged. The whole loop is maybe fifteen minutes per product once you've done it twice.
If you want to see how the dials behave before you commit to the process, the image to image tool will show you the difference between a light touch and an overworked pass in about three runs. Start at the low end, compare at full resolution, and let the reference do the work instead of the strength setting.