AI product photography — generating a product on a new background or in a new scene from a few reference photos — genuinely works for simple, rigid products shot cleanly to start with. It struggles with fabric, glass, reflective metal, and anything where the product's exact texture is part of what's being sold.
When AI product photography wins
- Simple hard-goods (bottles, boxes, hardware) that were already photographed cleanly.
- Lifestyle scene variations — the same product shown in ten different settings without ten different shoots.
- Fast turnaround for marketplace listings where speed matters more than perfection.
When a real shoot still wins
Fashion and fabric are the clearest case — drape, texture and how a garment moves on a body are exactly what AI generation still gets wrong. Food is another: the appeal is entirely in specific, real texture and light, and AI food imagery reads as artificial almost immediately to anyone who's actually eaten. And anywhere trust is the product — jewellery, cosmetics claiming a specific finish, anything premium — a real photograph signals real craft in a way buyers register even if they can't articulate why.
The approach that actually works for most brands
Shoot the hero images for real — the ones on your product page and top ads. Use AI generation for the long tail: seasonal variations, secondary angles, background swaps for a campaign refresh. That split gets the trust benefit where it matters and the speed benefit everywhere else.
We run both in the same production — a proper shoot for what needs to be trusted, AI-assisted variation for what needs to scale.