Flat lay to on-model AI: input, generation and review workflow (2026)

You can turn a flat-lay garment photo into an AI on-model image by supplying a clear view of the real item, choosing a garment-generation workflow and checking the redrawn result against the original. Treat the output as an illustration of the garment being worn, not proof of its real fit, hidden construction or condition.

What changes when a flat lay becomes an on-model image?

A flat lay records the garment resting on a surface. An on-model generation must create a person and a worn appearance: folds, contact with the body, sleeve position and surrounding scene. Those are additional visual decisions, not measurements recovered from a photograph.

Shopify’s clothing photography guide describes flat lay as photographing clothing from directly above and treats model, mannequin and lifestyle photography as distinct approaches. Keep those distinctions when adding AI. A plausible generated fold does not establish how the real fabric drapes on a real customer.

This is a practical workflow, not a report of a controlled comparison or a promise that one tool preserves every garment. The examples below are illustrative instructions. They are not customer results or measured performance claims.

The first decision is whether you need a worn illustration at all. If the job is to document a used item’s condition or show its actual measurements, an original flat lay may be the stronger deliverable.

Which input should I use for the image I need?

Choose the source according to the information the final image needs. A generator cannot verify a feature that is absent from the reference. Do not ask it to resolve uncertainty by making an attractive guess.

This is a source-planning table, not a promise that every tool accepts multiple reference images or can construct every requested view. Keep additional originals for review even when a workflow takes one product photo at a time.

CatalogCut’s flat-lay-to-on-model workflow starts from the garment photograph. If you need a conventional photographed fit record or construction evidence, arrange that photograph rather than substituting an invented view.

How should I photograph the garment before uploading?

Lay out the actual size and colorway you want to show. Use a plain surface that makes its boundaries readable. Arrange sleeves and hems so the garment’s shape is visible without overlapping another item. Leave space around the edges instead of cropping tightly through them.

Use consistent light and inspect the file for blur, glare and color casts. A window can be a useful starting point; avoid mixing visibly different light colors across the garment. Shopify’s guide recommends preparing clothing, planning angles and close-ups, and considering a neutral backdrop and natural light when setting up a shoot.

For a clean new garment, remove loose lint and prepare it as you would for ordinary product photography, following its care instructions. For used stock, preserve an honest record of wear and repairs. Do not digitally erase a condition feature that the buyer needs to see.

Before uploading, confirm:

• The whole garment is visible, with no missing cuff, strap or hem.
• The photograph is sharp enough to inspect its important details.
• Color, label and size correspond to the actual item.
• Prints, fastenings and distinctive seams are not hidden by folds.
• Hands, hangers or other garments are not obscuring required features.
• Original front, back, label, detail and condition files are saved separately.
• You have the rights to the photograph and any included person’s image.

The phone product photography guide covers the shooting setup. A better reference is useful even if you ultimately decide not to generate anything.

How do I make an on-model image in CatalogCut?

Upload the clear product photograph to Stage, choose the on-model shot type and use Creative mode. Set a simple scene and frame suited to the intended use, review the shown credit cost, then generate. Do not start with a complicated pose that hides the details you need to evaluate.

Creative redraws the whole image, including the garment. Preserve keeps the photographed product pixels and builds a scene around that photograph; it does not turn a flat garment into a newly worn view. How Stage works explains those modes and the source comparison before delivery.

Every Stage result is checked against the original before delivery. If it fails that check, its credit is refunded automatically. You should still compare accepted results yourself: a passed check is not certification that every logo, seam, material or fit implication is correct.

Standard output is 1K and one completed standard image uses one credit. Free accounts receive one Stage image per month with a small catalogcut.com mark. Basic is $7.99 per month for 75 Stage images, Pro is $17.99 for 180, and Studio is $34.99 for 350; paid plans remove the mark. Check the current plans for the allowance you actually need.

Stage does not generate video or upload to a store integration. Download the result, review it and upload it yourself only where the intended image is permitted. Check resolution needs before making a collection around standard 1K output.

What should I put in the prompt?

Describe the desired visible shot rather than asking the system to prove a physical fact. Specify the front or back view supported by the source, an uncomplicated pose, a plain setting and which product details must remain inspectable.

For an illustrative example, a brief for a front photograph of a navy cardigan might be:

“Front-facing adult model in a plain studio setting. Keep the cardigan fully visible with arms relaxed and no bag or outer layer covering it. Preserve the visible button arrangement, neckline, sleeve length and navy color from the source. Do not add decorative trims or accessories.”

That wording is a direction, not a guarantee. If the result changes a button, covers a cuff or invents a pocket, reject it even though the prompt requested otherwise.

Avoid writing “fits a size 12 perfectly” or assigning a real height and body measurements to a generated person. Those statements sound like fitting evidence. Use the actual garment’s size label and measurements for product information, and keep styling direction separate.

A reusable product photo style guide can define background, framing and light for a collection. Consistent instructions help you evaluate consistency; they do not establish identical results across every garment.

What should I inspect before keeping a result?

Compare the generated image, original file and garment together. Review the product before the face or scenery. A compelling model should not distract from a changed item.

Check silhouette, neckline, sleeve length, hem, pockets, closures, button count, print placement, logo spelling and visible texture. Inspect the boundary between the garment and skin or hair. Reject added straps, missing panels, false seams or an altered opening.

For an illustrative striped-shirt example, check whether stripes end at the correct seam and whether the chest pocket still exists. For a printed tee, read the actual lettering rather than accepting something that resembles a logo from a distance. These are review examples, not claims about failure frequency.

If the important detail is too small or obscured to verify, use a clearer source, request a simpler composition or keep the original photograph. Do not resolve an uncertain output by describing it as “close enough” while publishing it as the actual item.

The need for review is not unique to CatalogCut. FASHN’s own product FAQ acknowledges that AI can get small product details wrong and tells users to compare with the real item. That supports careful inspection, not a measured ranking between providers.

Can an AI model image show fit or replace a ghost mannequin?

It can illustrate a possible worn appearance; it cannot prove how a particular size fits a real person. A flat-lay photograph does not supply a physical fitting session, and a generated model is not a measurement record.

Keep garment measurements and actual size information available. If you need documented fit, photograph the real garment on an appropriate real model or mannequin and record the relevant details. Do not invent a generated model’s clothing size and imply it was tested.

A hollow ghost-mannequin photograph is another distinct deliverable. CatalogCut does not make true hollow ghost-mannequin output. Choosing an on-model scene does not replace that construction view. Decide whether the required result is a styling illustration, actual fit photograph or hollow garment image before spending credits.

Where can I use the finished image?

Check the particular destination, product category and account’s country before uploading. Owning the source and having commercial rights from a generator do not override a marketplace’s original-item-photo rules.

For example, Depop’s photo guidance tells sellers to use photographs taken by themselves. Vinted’s US catalog rules require seller-taken photos of the actual item and prohibit image editing. Neither requirement is solved by placing a synthetic picture last or labelling it as AI.

Poshmark’s guidelines prohibit misleading behavior and require respect for content rights, but that general language is not blanket preapproval of generated listing photographs. Keep original documentary shots available wherever you sell.

Where an illustrative image is permitted, make its role clear and retain any provider-required provenance during editing and export. Canva’s AI terms, for example, prohibit removing or disabling provenance metadata. Re-adding a plain disclosure field does not recreate a removed signed credential.

What to do next

Choose one garment with clear original photographs and define the exact image you need before generating. Review the AI clothing model workflow, then compare any result with the item and confirm its intended destination allows that kind of illustration.