The problem
Our client distributes color paint protection film: over 350 colors and finishes, from gloss Ferrari red to matte army green. Most of those colors sat on the website as drafts with no photos, because photographing every color on a real car would mean renting vehicles, wrapping panels, and shooting for weeks. You can't sell a car wrap nobody can see, so most of the catalog simply wasn't for sale.
What we built
An AI image pipeline that turns one supplier photo of a film roll into a full product listing: a cleaned-up roll shot, plus consistent car photos showing that exact color on a vehicle, from multiple angles, in studio lighting.
The hard part isn't generating one pretty picture. It's generating hundreds that look like they belong to the same store: same lighting, same angles, same color accuracy, with logos and text handled correctly. We built the pipeline with quality checks and cost tracking on every image, so each color costs cents to produce instead of hundreds of dollars, and a bad render gets caught before it reaches the store.
Results
- 93 previously unsellable products went live in the first wave, each with a full image set.
- Three orders and about $4,500 in sales came from those products within the first two weeks. Colors that had never been purchasable started selling almost immediately.
- New colors now go from supplier photo to published product in minutes, not weeks.