ChatGPT Prompts for Flat-Lay and Ghost-Mannequin Shirt Images

Use flat-lay prompts when you want an arranged garment view, and ghost-mannequin prompts when you want an invisible-support presentation with volume. In either case, keep the actual shirt reference central. A change in presentation must not quietly become a change in construction.

The main challenge is the information that a photograph does not show. Lifting a flat shirt into a three-dimensional-looking shape can cause the generator to invent an inner collar, lining or back detail. Treat those areas as uncertain until supported by additional references.

The images in this guide are AI-generated illustrations, not photographs of an item for sale or proof that a generated garment matches a real shirt.

Use references appropriate to the presentation

Prepare a full front view and close-ups of collar, cuffs, placket and hem. For a hollow-neck or inside-collar view, add a real interior reference when accuracy matters. Keep labels and logos out unless you have approved artwork and permission to use them.

ChatGPT Images supports creating and editing images through conversation; see OpenAI’s image guidance. The prompts below describe separate visual tasks rather than a guaranteed conversion process.

Eight copyable ChatGPT shirt-presentation prompts

1. Make a clean front flat lay

Create a front flat-lay concept from the supplied shirt reference. Arrange the garment on a neutral surface with sleeves relaxed and the complete hem visible. Preserve the documented collar, pockets, buttons and artwork. Use soft overhead light with gentle edge shadows and no props.

2. Keep cuffs and collar readable

Refine the flat-lay arrangement so both cuffs, collar points and upper placket can be inspected. Do not fold the sleeves across the shirt body. Preserve the reference construction and natural cloth outline. Avoid making the garment unnaturally rigid or perfectly mirrored where it is asymmetric.

3. Create a folded editorial alternative

Create a neatly folded shirt concept using the same reference, with the collar and a small section of the placket visible. Keep visible design details consistent. Do not add packaging, labels or accessories. This is an editorial presentation, not a replacement for the full garment view.

4. Explore an invisible-support silhouette

Create a ghost-mannequin-style shirt concept from the documented garment views. Show natural torso volume without a visible mannequin or person. Preserve visible construction and proportions. Do not invent inner labels, lining or hidden seams. Treat unsupported interior areas as concept-only details.

5. Use an actual inner-collar reference

Revise the inside-collar area using the supplied interior close-up. Match only the visible shape and construction in that reference. Keep the exterior shirt unchanged. Do not add branding, stitching or labels absent from the source, and do not present an uncertain detail as verified.

6. Ground the garment without an exaggerated shadow

Adjust only the studio shadow beneath the garment concept. Use a subtle shadow consistent with the light direction and presentation style. Remove duplicated shadows and floating-looking gaps. Preserve the shirt’s colour, outline, hem, cuffs and visible texture.

7. Preserve a documented back view

Create a separate back presentation using the actual back reference. Keep yoke, seams, pleats and hem faithful to what is visible. Do not copy the front’s placket or pocket onto the back. Keep the studio background and lighting direction related to the approved front frame.

8. Compare presentations for product drift

Compare the flat-lay and invisible-support candidates with the original shirt references. Check collar shape, visible button count, pockets, sleeves, hem and artwork. Distinguish perspective changes from construction changes. List uncertain interior details and do not approve either image automatically.

Keep the image’s job clear

A folded view may communicate styling but hide useful construction. A hollow-neck presentation may clarify volume but invent unseen information. Keep those trade-offs explicit and include real supporting views when the image represents merchandise.

For related workflows, read TryBuy’s flat-lay-to-model guide and reference-preservation prompts.

Frequently asked questions

Can a flat-lay image prove how a shirt fits?

No. Neither a flat arrangement nor generated volume establishes measured fit on a person.

Can I let the AI invent an inner label?

Not for a real product image. Supply approved artwork or keep the label absent rather than fabricating branding or care information.

Which presentation is better for ecommerce?

Choose according to the information the customer needs and the applicable channel requirements. A useful gallery can combine complementary views instead of forcing one image to explain everything.

Platform reference checked 26 September 2026. Independent TryBuy guide; not an OpenAI endorsement or accuracy guarantee.