Texture that reads as texture. A cream has to look like it would feel like something, a powder has to look like it would sit on skin, and a gloss has to look wet. That is the entire job in beauty, and it is the thing most product photography misses.
Shade ranges that stay true across a line, so a customer choosing between twelve colours online is comparing the product rather than comparing twelve different lighting setups.
Beauty is bought on a promise about how something will feel and look on the person buying it. Nobody can test that through a screen, so the photograph is carrying the promise on its own.
Shade accuracy is also a returns problem. Every customer who receives a colour that does not match the image is a refund, a restock and usually a review that mentions it. Getting the colour right in the picture is cheaper than getting it wrong at scale.
Generated beauty imagery invents texture, and texture is exactly what the customer is judging. A cream that never existed has a surface that behaves like nothing in the world, and people notice without being able to name it.
Fractl research in 2026 found the share of consumers saying heavy AI use would reduce their trust in a brand they liked doubled in a single year, from 20 percent to 40 percent.
Maybelline. Maybelline New York. GlamGlow Megamud and Instamud. Morphe. Murad. Olay, with Grey Hong Kong.
Related reading
Maybelline. Maybelline New York. GlamGlow Megamud and Instamud. Morphe. Murad. Olay, with Grey Hong Kong.