How AI is changing commercial product photography (and what it cannot replace)

There's a story about AI in photography that tech companies love to tell: the robots are coming, learn to prompt or get left behind. It makes for great marketing and terrible reality. I've been shooting commercial product photography for brands across Dehradun and Uttarakhand for years, and over the last twelve months I've pushed AI into every stage of my workflow — pre-production through final delivery. Not because someone told me to. Because the economics made it stupid not to. Here's what I actually learned, and it's messier than either the hype or the panic suggests.

The background problem AI actually solved

Before AI, every product shoot started with a painful question: build a set, find a location, or shoot on white and composite later. All three options cost real money. A kitchen set for a food brand could run you fifty thousand rupees or more. Location scouting meant permits, travel, and praying for decent weather. Even the white-background route needed a skilled retoucher to composite the product into a new scene without it looking pasted on.

Midjourney v6, DALL-E 3, Adobe Firefly — these changed the math. Not because they're perfect. They're not. But because they're good enough for eighty percent of the e-commerce and social work that crosses my desk. Shoot the product on a clean surface with controlled lighting, and you can drop it into a generated environment that actually looks photographic. The kitchen, the marble countertop, the lifestyle scene — all of it built in post.

The catch is lighting direction. If your product is lit from the upper left, the background has to honor that, or your eye catches the mismatch instantly. Most AI tools won't respect this on their own. You either prompt with ridiculous specificity or generate a stack of options and composite the best match. It's faster than building a set — it is not click-and-done.

And there are products where AI backgrounds still fall flat. Reflective and transparent stuff is the killer. A chrome faucet in a generated kitchen has reflections that never match, and no amount of prompting fixes that. A glass bottle with liquid inside breaks the illusion the second you look at it. Those still go on a real set. But matte, textured, non-reflective products — which is honestly most of e-commerce — the AI handles beautifully.

Automated retouching: the real productivity gain

If background generation is the flashy headline, automated retouching is the part that actually changed my business.

A typical e-commerce shoot gives me three to five hundred frames. Dust removal, color correction, keeping the batch consistent, cleaning up small imperfections — that used to take my retouching team two to three days. Now one person does it in four to five hours.

My daily tools: Retouch4me for batch dust and reflection cleanup, Luminar Neo for AI masking and sky replacement on lifestyle shots, and Photoshop's Generative Fill for pulling distractions out and extending backgrounds. None of them are perfect. All of them are good enough to eat the repetitive work that used to eat my post-production budget.

Here's the line worth understanding: cleanup retouching versus creative retouching. AI crushes cleanup — dust, color casts, exposure consistency across a batch. It still struggles with the creative side — keeping the skin texture that makes a lifestyle shot feel premium, preserving the small imperfections in a handmade product that tell a craft story, knowing when a reflection adds character and when it's just noise.

The workflow that works: AI does the first pass on everything. Then a human looks at every frame. That review takes about ten percent of the time manual retouching used to. You keep the eighty percent the AI got right and fix the twenty percent it didn't. That's not a theory — that's my actual pipeline.

The judgment problem nobody talks about

Here's the uncomfortable truth nobody in the marketing is telling you: AI is getting brutally good at execution and stays stubbornly bad at judgment.

A generative model can hand you a thousand background options. It can't tell you which one fits the brand. An AI retoucher can smooth skin until it's plastic-perfect. It can't tell you whether this particular brand wants perfect skin or real skin. An AI grader can match a reference look across a whole batch. It can't tell you whether that reference look is telling the right story.

And I don't think that's a temporary limitation. It's fundamental. AI learns patterns from data. Brand positioning, emotional resonance, cultural context — those aren't patterns in a training set. They're human decisions that require understanding the product, the audience, and the market in ways no dataset captures.

For photographers, this is actually good news. The ones who win are neither the Luddites nor the surrenderers. They're the ones who use AI as a production accelerator and keep creative judgment firmly human. Shoot with technical precision, let AI handle the mechanical work, then bring human judgment to the decisions that actually decide whether the image sells.

What this means for brands in 2026

For brands, the practical difference is real. Turnaround times drop — what took a week now takes two or three days. Costs drop — fewer set builds, smaller retouching teams. And iteration gets cheap — generating five background directions instead of one costs almost nothing, so creative teams explore more before committing.

But there's a catch, and it's a big one. The barrier to producing decent-looking product imagery has basically collapsed. Any brand with a phone and a Midjourney subscription can make images that look acceptable. That means basic production quality is no longer a differentiator — it's table stakes. What separates commercial photography from content-mill output now is the creative decisions: the lighting, the composition, the retouching judgment, the thousand small choices that say something about the brand.

For brands in Dehradun and across Uttarakhand, that's an opportunity. AI has democratized production quality. It hasn't democratized creative expertise. A photographer who knows both the AI tools and the traditional craft can deliver national-campaign quality at local production costs. That's the value proposition I'm building my work around.

The tools are available to everyone. The judgment to know when to trust them and when to overrule them — that's still rare. That's where the value lives.