The question comes up in every 2026 marketing meeting: "can't we just do the product photos with AI?". The short answer is: partly yes, and you save a lot — but not on everything, and not without some traps. The long answer deserves this article, because it's exactly the ground where we work with our AI Shooting service, and we've learned where AI shines and where it does damage.

Let's say it upfront, for honesty: we also sell AI-generated image production. Precisely for this reason, we won't tell you that AI replaces the photo studio. We'll tell you where each of the two wins, because that's how you do serious work — and because a customer disappointed by a wrong image costs more than what they saved.

Costs and times: an order-of-magnitude difference

The difference between the two approaches isn't measured in a few percentage points, but in orders of magnitude — and concerns both costs and times. Traditional shooting entails costs that grow with the number of images, because each shot brings photographer, studio, styling, and post-production. AI production starts from a low monthly fee and keeps a low per-image cost, which stays that way even when the images become hundreds. On timing the distance is just as clear: for a shooting, weeks are usually needed between briefing, set, and retouching; with AI you reason in hours.

More than the precise figures — which depend on the product, complexity, and vendor — what matters is understanding where each approach is efficient. The scheme below summarizes the main differences.

Cost per image — higher, grows with volume · low, marginal on large numbers

Times — weeks (briefing, set, post-production) · hours

Large catalogs — cost proportional to shots · scales without proportional costs

Variants and tests — new set, new cost · immediate, at minimal cost

Where AI wins (and deserves to win)

There are jobs where AI is not just cheaper: it's simply the right tool. Backgrounds and lifestyle settings are generated in minutes instead of setting up a physical set. Color or size variants of the same product are produced in series, consistent with each other, starting from a single template. Scalability is its superpower: updating a catalog of hundreds of items has a cost and time that traditional can't come close to. And there's a use often ignored but valuable — pre-visualization: seeing how a product will render in a certain context before organizing (and paying for) the physical shooting.

Where AI gets it wrong (and here it's no joke)

This is the part that those who only sell AI tend to skip. Yet it's what separates professional work from an embarrassment.

Text on packaging and labels. It's the most serious and systematic limit. Models treat writing as texture, not as words: they produce typos ("Vitmain D3"), altered brand names, illegible ingredients, invented characters. For any image where the label must be readable — food, cosmetics, pharmaceutical — AI alone isn't an option. Text must be applied in post-production, separately.

Reflective materials, glass, transparencies. Glass, polished metal, and transparent packaging are the Achilles' heel: AI doesn't correctly replicate the physics of light, and it shows.

Logos and visual identity. When a product is regenerated in different contexts, logo and typography can "drift" subtly — the so-called visual drift. For a brand this is a consistency risk, not just aesthetic.

Jewelry and fine details. Along with packaging with small text, it's the category where AI struggles most. And these are often high-margin products, the ones where a mediocre image costs the most.

Operational translation: AI is not suitable for the main image (the hero image on the product page) of your flagship items, nor for packaging with regulatory information. There you still need the camera.

The legal knot, new and often ignored

There's a chapter that in 2026 can no longer be brushed off: AI-generated images have transparency obligations. From August 2, 2026, article 50 of the European AI Act comes into force, with a two-level system. On one side, the tool that generates the image must embed a "machine-readable" technical marker (industry standards, such as C2PA, are already heading in this direction). On the other, whoever publishes the image — the brand or agency — must declare when content is artificially generated, in particular if it realistically depicts people, places, or products. The provided sanctions aren't symbolic: they reach €15 million or 3% of worldwide turnover.

To this is added a rule that already applies today, AI or no AI: the Consumer Code (and the European directive on unfair commercial practices) prohibits images that mislead about the actual product — color, size, features. A generated image that makes the product look different from what it is may constitute an unfair practice, with AGCM sanctions up to €5 million. Finally, a warning on rights: in Europe an image purely generated by AI, without substantial human creative input, generally isn't protected by copyright. If the image is a valuable asset, you should know this.

The truth: the right answer is "both"

Once costs, limits, and regulations are lined up, the conclusion isn't "AI yes" or "AI no," but a hybrid approach — and it's what we recommend almost always. Traditional shooting for what really matters: the hero image, flagship products, packaging with text, difficult materials. AI for everything else: color variants, lifestyle settings, seasonal and social content, visual tests. Plus, AI pre-visualization before the physical set reduces "blind" shoots and thus the shooting costs themselves. The typical result is a 40–60% saving on the overall photo budget, without losing quality where quality is noticed.

Who's already doing it

This isn't insider theory. As of June 2026 Amazon has started inserting AI-generated lifestyle images in mobile search results (for visual discovery, not as photos of purchasable products). IKEA, with IKEA Kreativ, uses AI and computer vision to show its products in the customer's real environment. They're different examples, but they tell the same story: generated imagery has entered the workflows of the big players, selectively and thoughtfully — exactly the opposite of "let's throw away the photographers."

In summary

With AI you save a lot, and on many images it's simply the right tool: wide catalogs, variants, lifestyle, tests, ultra-fast times. But on hero images, packaging with text, reflective materials, and premium products, traditional photography remains irreplaceable — and from 2026 there's a legal transparency layer to respect. The smart move isn't to pick a side: it's to combine the two, putting each where it delivers the most.

Sources

Data and prices updated as of June 2026. Figures on traditional shooting costs largely come from AI solution vendors and should be read as orders of magnitude. This article has informational purposes and does not constitute legal advice.