A studio product shoot used to mean a booking, a photographer, a lighting rig, and a week of waiting on edits. Now you can describe the shot and get it back in a couple of minutes. This is a working playbook for using AI product photography to make ecommerce photos that actually sell -- clean pack shots, on-model looks, seasonal scenes -- without renting a single softbox.
What AI product photography actually does well (and where it doesn't)
Let's be precise, because there's a lot of hype. AI is excellent at context -- putting a product you already own into a new scene, on a new surface, in new light. It's weaker at inventing a product from scratch that has to match a real SKU exactly. So the winning move for most stores isn't "generate a product from a text prompt." It's product recontext: you hand it a real photo of your real item, and it re-lights and re-stages it while keeping the actual object intact.
That distinction matters for one reason above all: your label, logo, and shape have to stay accurate. A hero image with a warped logo or a misspelled ingredient list isn't a nice image -- it's a returns magnet and a trust problem. Good tools lock the product and change only the world around it.
The four shots every product page needs
Before you generate anything, decide what you're actually making. Most stores over-index on one glamour shot and skip the images that do the quiet conversion work. Aim for a set:
- The clean pack shot. Pure white or soft neutral background, even light, no distractions. This is your grid thumbnail and marketplace-compliant hero.
- The in-context lifestyle shot. The candle on a linen bedside table at dusk. The serum on a wet bathroom shelf. This is where AI product images earn their keep -- you can shoot ten "locations" without leaving your desk.
- The scale-and-detail shot. A hand holding it, or a tight macro on texture and stitching. Buyers zoom. Give them something to zoom into.
- The seasonal or campaign shot. The same product in autumn tones, holiday reds, or a summer picnic. Regenerate this every quarter instead of re-shooting.
Write prompts like an art director, not a search engine
The single biggest quality jump comes from specificity. "Product on a table" gets you slop. Direct the shot the way you'd brief a photographer. Name four things every time:
- Surface and setting: "on weathered oak," "on a wet marble ledge," "on cracked desert clay."
- Light: "soft window light from the left, late afternoon," or "hard studio key with a subtle rim light."
- Camera feel: "shot at 85mm, shallow depth of field, product tack-sharp, background falling off soft."
- Mood and palette: "warm, quiet, editorial -- muted earth tones," not "beautiful, amazing, professional."
Adjectives like "stunning" and "high quality" do nothing. Nouns and physical details do everything. If you can picture the exact photo in your head, the model can too.
A quick before/after
Weak: "My coffee bag, nice background, professional."
Strong: "My coffee bag standing on a rough concrete counter, morning light raking from the right, a blurred cafe interior behind it, 50mm, shallow depth, warm and grainy, product label sharp and fully readable."
Same product. Wildly different result. The second one reads like a real photo because you gave it real photographic decisions to make.
Keep the product honest
This is the rule that separates ecommerce AI photos you can actually ship from ones that get you a chargeback. A few guardrails:
- Start from a real product photo whenever the exact object matters. Recontexting a true image beats generating a lookalike every time.
- Zoom in and audit the label before you publish. Check spelling, logo geometry, and any legally required text. If it's warped, regenerate -- don't ship it.
- Match your brand palette on purpose. If your brand is cool and minimal, don't let the model hand you warm rustic every time. State the palette, or lock it with brand tooling so every image looks like it came from the same store.
- Watch the physics. Reflections, shadows, and the way a bottle sits on a surface are the first tells of a fake. A shadow going the wrong way is more damaging than a slightly boring composition.
A repeatable workflow you can run this afternoon
Here's the loop I use to go from one phone photo to a full product page's worth of images:
- Shoot one clean reference. Even a decent phone photo on a plain background gives the model a truthful starting point.
- Recontext into your four shots. Run the pack shot, the lifestyle scene, the detail crop, and the seasonal look as separate prompts.
- Generate three variants of each and pick the winner. Variants are cheap; indecision is expensive. Judge on realism and label accuracy first, prettiness second.
- Upscale the keeper to the resolution your store needs, then export at the aspect ratios each channel wants -- square for the grid, vertical for stories, wide for the banner.
- Log what converted. When a lifestyle shot beats a pack shot in the wild, note the setting and reuse it for your next SKU.
Do this once and you'll never dread a new product drop again. The marginal cost of "same item, new scene" drops to near zero, which means you can test image angles the way you test ad copy.
Where Bloopo fits
Everything above is exactly what Bloopo is built to do. It's a connector you add to Claude or ChatGPT -- point them at mcp.bloopo.ai/mcp -- and then you just ask, in plain language, for the shot you want. It runs top image models under the hood, does true product recontext that keeps your labels and shapes accurate, and has an anti-slop layer that catches the warped-logo, wrong-shadow tells before they reach you. You see the price before anything runs, and a batch of product images costs cents, not a shoot day.
If you've got one honest photo of your product and five minutes, that's enough to start. Add Bloopo in the chat you already use and ask it for your first pack shot.