Product photography at scale: what breaks first

It shows up somewhere after product two hundred. Not as one bad image, but as a page that feels slightly restless. Shadows fall two ways. Three whites are in use. Half the range is photographed at eye level and half slightly from above. Nothing is wrong enough to reject, and together it reads as a shop that does not quite have its act together.

This article is about where that drift actually comes from, and what to fix first. It is written for brands with a range that keeps growing and images that end up in places they do not control.

The four moments where consistency breaks

In our own production we see the same four moments, in roughly this order.

A new category arrives. You had ceramics, now you also have textiles. The rules you wrote for a hard product with clean edges do not survive contact with something soft that drapes. Someone improvises, and the improvisation becomes the new standard for that category.

Someone new does the work. A second photographer, a new agency, a colleague who took over the task. They read the same brief and reach a different conclusion, because the brief described a feeling rather than a setting.

A launch is in a hurry. The range goes live on Monday, the images arrive on Friday, and the shortcut is a one-off. Six months later the one-off is thirty products deep.

A second market opens. Different packaging, a different plug, a different language on the label, a retailer with its own image specification. Suddenly there are two versions of the truth and no rule about which one is the real one.

Note what these have in common: none of them is a quality problem. Every individual image is fine. The problem is that the decisions behind them were never written down in a form anyone could follow.

Consistency is a set of decisions, not a style

"On-brand" is not a usable instruction. This is, and it fits on one page:

Background. The exact value, not "white". Whether there is a shadow, and what kind.

Light direction. One direction for the whole range, and how hard the shadow is allowed to be.

Angle set. Which angles every product gets, in which order. Front, three-quarter, detail, in use.

Framing. How much space around the product, and which aspect ratios you deliver.

Scale. How a viewer can tell how big the thing is. For furniture and sanitary ware this is the single most returned-for reason.

Colour reference. What the product colour is measured against, and who decides when a render is off.

Retouching. What gets removed and what stays. A real material has texture; a perfectly smooth one looks fake.

Naming. How files are named, because the people downstream search on exactly that.

Write these as a reference set, not as prose. One approved image per rule beats a paragraph describing it. Everyone who makes images for you, inside or outside your company, should be able to hold their result next to your reference and see the answer immediately.

The rule nobody writes down: who owns the rules

Here is the part that is missing from most style guides we are shown. The document exists, it is good, and it is six versions behind, because the first exception was granted by someone who no longer works there.

Name an owner. One person decides what an exception is, and every exception is added to the reference set or refused. Without that, your guide degrades into an archive of what you used to think.

What AI changes, and what it does not

Generating images at volume removes the production bottleneck. It does not remove the consistency problem; it just moves it earlier.

What genuinely gets easier:

• •The tenth, fiftieth and two hundredth image can come out of the same setup, because the setup is a saved instruction rather than a room you have to rebuild.
• •A new colour or a new variant does not mean a new session.
• •A second market gets its own version of the same scene without a second shoot.

What does not change:

• •If your rules are vague, AI produces vague output faster.
• •Materials still need judgement. Brushed metal, glass, woven fabric and anything transparent are where generated images give themselves away, and where a person has to look before it goes out.
• •Approval still has to happen somewhere. Two hundred images a week is only an advantage if two hundred images a week can be checked.

That is why we keep a person in the loop on everything we deliver. Not as a philosophy, but because we have seen the alternative at volume.

The layer most articles skip: your images are used by other people

If you sell through dealers, retailers and marketplaces, consistency is not something you can verify on your own website. Your range appears on pages you do not control, mixed in with competitors who are not following your rules at all.

Three things go wrong there, every time:

1. The partner publishes the easiest file. That is usually the packshot, because it is the one that fits their template. Your lifestyle photography never arrives on their page.

2. The partner uses what they already have. Last year's set, from an expired link, because the new one was hard to find.

3. Each market drifts on its own. The German dealer needs German packaging in shot, the Danish retailer has its own crop specification, and one marketplace compresses everything you upload.

So the practical test of consistency is not your own product page. It is the search result for your own product name, on the pages you do not own. Run it once. It is uncomfortable and it tells you more than an audit of your own site.

What this looks like in practice

Two examples from our use cases.

Xenz | Beterbad Sanitair makes baths and taps. We produced print-ready bathroom scenes from their technical drawings, within three days. The three days are the part people remember, but the useful part came afterwards: a route that every new product goes through, so the next product arrives in the same style without anyone deciding again.

DHK Kozijnen sells windows and doors. Instead of showing products in isolation, we placed them in realistic Dutch homes. The images worked, but they only worked at scale once the whole range had the same treatment and looked the same everywhere it appeared.

In both cases, the value was not the first image. It was the fiftieth looking like the first.

A workable order

1. Fix the rules on your best sellers first. Not on your archive. The archive can wait, possibly forever.

2. Build a reference set, one approved image per rule. Share it with everyone who produces for you.

3. Decide what "complete" means per product. Which images, which formats, which markets. Write the list once; it is the same list every time. If that line is not drawn yet, start with which image each channel needs.

4. Name the owner of the rules. Exceptions go through one person.

5. Check the outside, not the inside. Once a quarter, look at your product on partner pages and marketplaces, and fix what you find at the source.

Step five is the one that usually finds the expensive problem.

Where LensLab fits

We built LensLab as one layer rather than three tools, for this exact reason. Our creators produce your visuals with AI, with human quality control through to delivery, based on the Brand DNA of your brand so the style holds across the range. Everything lands in one central library, and through the distribution module your dealers, retailers and sales partners get one door that always holds the current set. You can see how it works in more detail.

To be straight about what is not there yet: Shopify and WooCommerce integrations are planned for November 2026, and insight into which images are actually being used is in development.

In closing

Consistency at scale is not a photography problem. It is a decision problem that photography happens to expose. Write the decisions down, give them an owner, and check the places you do not control.

Want to see what that looks like for your range? Try LensLab free with five credits, or have a look at what we made for other brands first.