Field Notes — Nº 02 · July 2026
AI Product Photos for Ecommerce
A PDP playbook for fashion retailers — how AI product photos, on-model variants and ecommerce photography systems actually earn their place next to a hero campaign.
The PDP is where taste meets conversion
Most fashion brands still budget imagery as if the campaign shoot were the point. It isn't. The campaign gets the headlines; the product detail page gets the credit card. On a typical retailer, seventy per cent of revenue moves through a page nobody put on the moodboard.
That is the page that gets the worst images — recycled lookbook crops, a single flat lay, one washed-out on-model shot in a size that isn't yours. AI product photos exist to fix exactly that problem.
What a complete PDP actually needs
A serious product detail page is not one image. It is a system: front, back, side, detail, on-model, in-context, and a scale reference. Six to eight images per SKU, minimum. Multiply by two thousand SKUs a season and the arithmetic explains why most PDPs are half-empty.
A traditional studio cannot fill that pipe. It was never designed to. AI ecommerce photography was.
Where AI product photos fit — and where they don't
Fit: PDP grids, colour and size variants, market-specific styling, refresh cycles on evergreen categories, and the long tail of SKUs the campaign photographer never had time for. Anywhere the customer needs a clean, consistent, trustworthy image of the actual garment.
Not fit: the campaign hero, the founder portrait, the story-driven editorial. AI is not here to replace the photographer whose eye is the reason the brand exists. It is here to stop asking that photographer to shoot a size-16 midi in nine colours at midnight.
The Comparison
I.
Studio Only
Great hero image, empty PDP. Long-tail SKUs never get shot. Reshoots on every colourway.
II.
Generic AI Tools
Cheap, fast, inconsistent. Faces drift between shots. Fabric physics is wrong. Fine for moodboards, wrong for a PDP.
III.
Parallel Studio
Consistent model identity across every SKU. Your aesthetic coded as a system. PDP-ready output at ecommerce scale.
The four-step PDP workflow
One. Upload the garment — flat, ghosted, or already on a fit model. Two. Choose the model archetype: age, body, market, mood. Three. Set the background system: studio grey, editorial location, brand-specific set. Four. Generate the full angle grid at scale, versioned rather than reshot.
What used to be a two-week production schedule collapses into an afternoon. What used to require a call sheet becomes a taste decision.
What "on-model" really means for ecommerce
A single perfect shot is not the goal. Consistency is. The customer clicking through fourteen dresses needs to read them as a coherent range — same model identity, same lighting language, same crop grammar. That is what turns a PDP grid into a wardrobe instead of a warehouse.
Consistency at that scale is only possible when the model is a system rather than a booking. That is the shift AI product photos actually deliver.
The honest recommendation
Keep the campaign in the studio. Move the PDP into a system. The maths tips somewhere around thirty SKUs and never tips back. Below that, taste is cheap. Above it, taste is infrastructure — and AI ecommerce photography is how you build it.