A generated image of your product is a picture of a product that does not exist, and a marketplace listing built on one is a misrepresentation dressed as efficiency. The useful version of a product photography AI prompt is a brief: a shot list written as precisely as a photographer would read it. That works with an image model as a reference, a photographer, a phone and a lightbox, and it never puts an untrue image on a listing.
The rule underneath everything here is simple. Generate backgrounds, surfaces and styling ideas if you like, but the product itself has to be photographed. Buyers comparing listings are looking at the object they will receive, and generated detail is where that comparison fails.
Why briefs beat generated product images
There are four reasons, and only one of them is about effort.
- Accuracy. Generated products invent a stitch pattern, a logo placement, a lid hinge, a colour. Buyers receive something else.
- Platform rules. Marketplaces require images to represent the actual item being sold. A generated product image sits in grey territory that can end with a listing removed.
- Returns. The returned item and the listing image do not match, which is the single most common avoidable reason for a return.
- Reuse. A brief documents what you shot, so the next product in the same range reuses the setup rather than starting from nothing.
Generated images still have a place. Backgrounds, colour swatches, lifestyle scenes with no product in them, and styling mood boards are all legitimate, and the brief approach makes those easier too, because you are specifying the scene rather than inventing the subject.
The shot list prompt: what to shoot
Start from the listing, not from a camera. Each marketplace category has an expected image set, and the brief should be written against it.
Write a product photo shot list for [product] being sold on [marketplace]. My setup is [phone or camera, lens if any, available light, a white sheet, a window, budget for one light]. Produce [number] shots, each on its own line, in this format: Shot [n] - [purpose in one phrase]. What to show: [what is in frame, in plain terms]. Camera position: [above, at eye level with the product, three-quarter, flat lay]. Background: [description]. Light: [where the light comes from]. Props: [only what I own]. Watch out: [the mistake most likely with this product]. Cover the main image first, then detail shots, then scale or dimension shots, then a use-in-context shot. Include at least one shot that shows the size of the product against a known object. Do not invent product features I have not described. Do not suggest props I said I do not have. Do not add retouching that changes colour or shape. If a shot needs a specific measurement, write [CONFIRM: measure before shooting].
Fill in these details
- Product: the actual object, with the variants that matter. A colour range needs one brief that names which shot changes with colour.
- Marketplace: the platform, because image requirements and main-image rules are set per platform and they change. Check the current listing requirements rather than trusting a brief.
- Setup: what you really have. A brief written for a studio is a brief you will not follow.
- Known dimensions: anything you have measured. Do not let the model estimate a size you will print on the listing.
The "watch out" line does more work than it appears to. It is the difference between a shot list and a wish list, and it is where the product's own awkwardness gets named before you set up.
Turning a brief into a reference prompt
Once the shots exist as a list, an image model can help you block a scene. Use it for everything except the product.
Using this shot list as reference: [paste the shot list] Describe, in one paragraph each, the staging for shots [numbers]: where the subject sits, the light direction and quality, the background treatment and the camera height. Then write one image generation prompt for each of those shots that produces the background and staging only. Describe the light, the surface and the environment in detail. Do not describe, generate or reference the product itself. Leave the subject area empty. Do not add text, logos, labels or hands.
That last paragraph is the whole technique. The model renders the table, the window light and the drape, and you place your real product into it. If a model refuses a prompt because you withheld the subject, that is a good sign the constraint is working.
Worked example: one kettle, seven shots
A seller with a stainless steel kettle, a window and a white sheet runs the first prompt and gets this.
| Shot | Purpose | Camera | Watch out |
|---|---|---|---|
| 1 | Main image | Three-quarter, eye level | Handle reflection blows out against a window |
| 2 | Lid hinge detail | Close, slightly above | Fingerprints show at this distance |
| 3 | Base and power socket | Flat, straight on | Cable runs out of frame |
| 4 | Scale against a mug | Side by side, eye level | Kettle looks smaller than it is without the mug |
| 5 | Capacity in use | Slightly above, in hand | Water level must be plausible for the fill line |
| 6 | Finish in daylight | Close on the steel | Chrome finish picks up the room, use plain backdrop |
| 7 | Box and included parts | Flat lay | Confirm what is actually in the box |
Four of those seven shots need nothing but the product on a surface. Only shot 5 involves handling, and it is the one where a generated image would be most tempting, which is exactly why the brief names the fill line.
Check the images before they go on the listing
- Compare image to object. Put the product next to the screen. Colours, proportions and markings should match what a buyer receives.
- Check the main image against platform rules. Main image requirements are specific per platform and change. Read the current listing image policy.
- Look for generated detail. Stitch patterns, hinge shapes, printed text and reflections are where generated products betray themselves.
- Confirm the scale shot is honest. If the listing says it holds 1.7 litres, the scale image should support that.
- Remove anything that changed the product. Generative fill that tidied a scratch, smoothed a dent or altered a colour is a misrepresentation, even when it looks better.
Mistakes that produce unbuyable listings
- Generating the product itself. The problem this whole approach exists to avoid.
- Writing the brief from the product name. A brief written without handling the object names shots you cannot take.
- Copying a category template. Ten identical flat lays across a range makes the range look like one product.
- Ignoring the marketplace's own image rules. They are per platform, they are specific, and they change.
- Letting AI retouch fix a bad photograph. Reshooting takes ten minutes and removes the risk entirely.
- Briefs without a scale shot. The size question is the top reason buyers return small items.
Frequently asked questions
Is AI-generated product photography allowed on marketplaces?
Generally the image must represent the item actually being sold, and a generated product is not that item. Each platform sets its own image and AI content rules, and they change, so read the current policy for the marketplace you list on.
What can I use an image model for instead?
Backgrounds, staging, surfaces, lighting references, styling directions and mood work. Anything where your product is not being depicted is a legitimate use and makes the real shoot faster.
How many images does a listing actually need?
Enough to answer the questions buyers ask before they buy, which is usually main image, detail, size and context. More images of the same flat angle add nothing and take time you could spend on the copy.
Should I use a real photographer?
For a product range you shoot repeatedly, yes, and the brief format is exactly what makes a photographer's work cheaper, because you arrive with the shot list decided. For a first product, a window and a white sheet will do.
Write the brief for the next product you list
Take the product you will list this week, set it on the table in front of you, and write the shot list by hand first. Then run the prompt to catch what you missed and add the watch-out lines you would have forgotten. For the craft behind the shots, browse photography and visual storytelling.