Use case · AI content creation

AI product photography at catalogue scale.

AI product photography works at catalogue scale when it is run as a standard rather than a series of one-off images: a fixed shot list per range, the same light and shadow on every SKU, and the real product locked so labels, proportions and colours never drift. Main images must still meet each marketplace's rules, which are stricter than most brand guidelines.

The shot list is the catalogue standard

Shoppers compare products side by side, so a catalogue has to behave like one set. Write the standard down per range: the angles (front, three-quarter, back, detail, in-hand for scale), camera height, margin around the product, light direction and shadow type. A long-lens look without wide-angle distortion keeps proportions honest. Decide whether the range uses soft drop shadows, hard cast shadows or none, and hold that across every SKU. Generation makes another angle cheap; it does not make it cheap to fix a thousand images produced to five slightly different standards. The standard is the brief, and every output is checked against it.

Why the real product has to be locked

Image generators redraw what they are shown. Left alone, they soften label text into lookalike letters, nudge a logo, add a button, round a corner or shift a colourway towards whatever they have seen most often. On a campaign image that is embarrassing; on a product detail page it misdescribes the goods. The dependable method starts from real pixels, a photograph, CAD render or 3D scan of the actual product, composited into generated light and surroundings rather than redrawn. Check every output against the reference at full resolution, paying most attention to small print, ingredient panels, sizes and marks such as CE or recycling symbols.

What marketplaces demand of main images

Amazon's main-image rules are the strictest most brands meet: a pure white background (RGB 255, 255, 255), the product filling 85% of the image, and no props that are not included, promotional claims or Amazon badges. The main image must be a realistic, professional-quality image that shows the product's real scale, quantity and colour, never a placeholder, and 1,000 pixels or more on the longest side enables zoom. Google Merchant Center rejects promotional overlays and watermarks, and asks merchants to keep the embedded metadata that marks an image as AI-generated. Lifestyle scenes, infographics and scale shots belong in the secondary images, where the rules are looser.

Retouch standards, applied to the whole range

Write retouching rules as tightly as the shot list: remove dust, fingerprints and packaging creases; straighten labels; match colour to the physical sample under controlled light. Never remove seams, stitching or features the customer will receive, and never change size, finish or texture. Then apply the rules to the range in one pass rather than image by image. In Synthetic White, products are saved as elements under product-lock, so geometry, labels and colours stay pixel-accurate while scenes change, and batches can run straight from a PIM or product feed. Approvals cover the batch, with an audit trail and a record of which model made each image.

Updated 25 September 2026

Questions

AI product photography, answered.

Can AI product photos be used as the main image on Amazon?

Amazon asks for a realistic, professional-quality image on pure white that shows the product's real scale, quantity and colour, so a fully generated render that drifts from the real product is a risky main image. A real photograph, cut out and lit to the range standard, is the safe route. Generated scenes suit the secondary images, where lifestyle, scale and detail shots are expected.

How do you keep colours accurate across thousands of SKUs?

Control the reference and the check, not just the output. Capture or render against a colour target, work in a managed colour space, and compare outputs with the physical sample under standard light using a measurable tolerance. Generated scenes can tint a product through reflected light, so check colour after compositing as well as before.

What does AI replace in a product photography workflow, and what not?

It replaces much of the repeat work: new backgrounds, seasonal scenes, extra angles from a 3D model and versions per market. It does not replace an accurate reference of the product. Every range still needs correct source photography or CAD data, because a generated scene is only as truthful as the product it starts from.

Get started

See your AI studio generating this week.

Thirty minutes. A live studio in your colours, your brand hub loaded, a real campaign brief.

See plans