Ghost mannequin photography, also called invisible mannequin photography, shows a garment as if someone is wearing it, but with no body or mannequin in the final image. You photograph the garment on a mannequin, photograph the inside of the neck and back separately, then remove the mannequin in editing and fill the gap with the inside shot. The result keeps the garment's 3D shape, which a flat lay cannot.
The short version: use a mannequin with removable neck and chest pieces, lock the camera and lights, shoot every view twice (on the mannequin, then the inner neck and back), composite them in Photoshop or a similar editor, and export a master per view. Check the channel rules before you shoot. Zalando rejects visible mannequins and wants the inner back view, and Amazon asks for adult clothing main images on a model, so ghost mannequin shots go into Amazon's extra images instead.
Ghost mannequin, flat lay or on-model: which garment photo for which job
Most apparel catalogs use all three styles. They differ in what they show and in how much setup each item needs.
- Ghost mannequin: shape, neckline and drape without a model. One person and a fixed setup can shoot a whole range in the same pose, so a category page looks uniform. It needs an editing step per image.
- Flat lay: the garment laid flat and styled from above. It is the quickest to shoot and edit, works well for kids' clothes, socks and accessories, but shows little fit or volume.
- On-model: fit, length and movement on a person. It sells fit best, but it needs models, a stylist and more retouching, and the model can distract from the garment.
A common split for webshops: ghost mannequin for the main catalog image and the back, on-model for one or two fit and styling shots, flat lay for accessories. Marketplaces then decide which of those may be the first image, as the channel table further down shows.
Mannequin, lights and camera for an invisible mannequin setup
The setup has one goal: every plate of a garment, and every garment in a range, is shot from exactly the same place with exactly the same light. Compositing only looks real when the two images match.
- Mannequin: a ghost mannequin with detachable neck, chest and arm pieces, in the size you sample in. White or light gray blends into a white background and makes masking easier. Use a lower-body form with a removable waist plate for trousers and skirts.
- Camera: on a tripod, at chest height for tops and hip height for bottoms, with a standard or short telephoto lens so the garment is not distorted. Mark the tripod and mannequin positions on the floor with tape.
- Settings: manual exposure, the camera's base ISO, an aperture around f/8 to f/11 so the whole garment is sharp, and a custom white balance taken from a gray card. Shoot tethered to a laptop so you can overlay the previous shot.
- Light: two softboxes at roughly 45 degrees either side, plus a separate light on the background if you want it pure white. Keep the power and position fixed for the whole session.
- Prep kit: steamer, lint roller, pins and clips to take in excess fabric at the back, tissue paper to fill sleeves, and tape to hold labels flat.
The ghost mannequin shot list per garment type
Each view needs a second image of the inside that the mannequin hid. Which plate that is depends on the garment.
Shots and inner plates per garment
| T-shirt, sweater | Front, back | Inside of the back neck with the label, shot with the neck piece removed or the garment inside out | Print, fabric, hem, cuff |
|---|---|---|---|
| Collared shirt | Front buttoned, back | Inside of the collar and yoke, collar opened flat | Buttons, cuff, fabric |
| Jacket, coat | Front closed, back | Inner neck and lining, hood folded down | Zip, pocket, lining, hood |
| Dress | Front, back | Inside of the neckline and strap tops | Fabric, closure, hem |
| Trousers, jeans | Front, back on a lower-body form | Inside of the waistband with the label | Waistband, pocket, hem |
| Swimwear, kids' underwear | Front, back | Inside of the back and straps | Fabric, lining |
Then add the shots shoppers use to judge the product: a fabric close-up, the closure, a pocket or print, and the care and size label if you are allowed to show it (our care label guide covers what that label has to say). For a full product photography routine beyond apparel (equipment, lighting and shot lists per category) see our product photography guide.
Editing the composite: from two plates to one garment
The editing is the same for every view: cut the garment out of the mannequin shot, place the inner plate behind it, and blend the join.
- Open both plates as layers in one file, mannequin shot on top, and add a white background layer at the bottom.
- Select the garment in the mannequin shot and mask out everything else, including the mannequin's neck, shoulders and any visible pins. Keep the mask, do not delete pixels.
- Mask the inner plate the same way, so only the inside of the neck or waistband remains.
- Move the inner plate behind the garment layer and scale or warp it until the neckline edges line up with the front.
- Add a soft shadow inside the opening so the inside reads as further away. Without it the neck looks flat.
- Retouch: remove lint, loose threads and creases that the steamer missed. Do not change the garment's shape, length or color.
- Check color against the physical garment or the color reference card from the shoot, under the same monitor settings for every image.
- Save a layered master at full resolution, then export per channel from that master. Never resize an already-resized copy.
Ghost mannequin editing workflow
Ghost mannequin rules on Amazon, Zalando, Google and Shopify
Channels disagree on two things: whether a mannequin shot may be the first image, and the size and ratio of the file. These are the rules as each channel published them on 3 November 2026.
Mannequin rules and image specs per channel
| Amazon (amazon.com) | No for adult clothing (standing model required); yes for extra images | In the main image, no part of a mannequin may show, including clear, flesh-toned or hanger. Children's and baby underwear, leotards and swimwear are shot lying flat, without a model, in every image | Main image on pure white, product filling 85% of the frame; clothing main images close to 3:4 portrait; 500 px minimum on the longest side, 1,000+ px for zoom (and for all children's wear), 10,000 px maximum |
|---|---|---|---|
| Zalando | Yes, as the primary packshot for apparel | Visible mannequins are rejected. Invisible mannequin or flat lay is required for kids' nightwear, underwear and beachwear. Always include the inner back view | Portrait 1:1.44; minimum 762 x 1100, best 1801 x 2600 px; JPG, sRGB, max 20 MB; 3 images for apparel |
| Google Merchant Center | Yes, no mannequin rule | No promotional overlays, borders or placeholders; images made with generative AI must keep their AI metadata | 500 x 500 px minimum for all products from 31 January 2027; about 1500 x 1500 recommended; max 64 MP and 16 MB |
| Shopify | Yes, the theme decides | No mannequin rule | Up to 5000 x 5000 px or 25 MP, under 20 MB; 2048 x 2048 usually displays best for square images; PNG, then JPEG |
Sources: Amazon's product image requirements and image guidelines for clothing, Zalando's image guidelines and apparel image guide, Google's image_link specification and Shopify's product media help. Zalando also gives white primary packshots a light gray background automatically, and lists visible mannequins as its own rejection reason (ACS 123). Our Zalando listing guide covers the attribute side of the same listing.
Format choice per channel (JPEG, PNG or WebP) is covered in WebP vs PNG for product images. For ghost mannequin masters, keep a lossless layered file and export JPEG for Zalando and Amazon.
Six ghost mannequin defects that get images rejected or returned
Most problems show up only when the image is viewed at full size, on the channel, next to the garment's other colors. Check every final image for these before upload.
- Mannequin edges left in: a sliver of neck form inside the collar or a shoulder edge under a thin fabric. On Zalando a visible mannequin is a rejection reason in itself.
- Mismatched inner plate: the inside of the neck shot from a different angle or under different light, so the opening looks pasted in or tilted.
- No inner back view: the neck opening filled with white or a flat gray shape. Zalando asks for the inner back view, and an empty hole reads as a cheap edit.
- Color drift between variants: the navy shirt looks black and the sand chino looks gray because each color was corrected by eye. Correct against the reference card, not against the previous image.
- Pins and clips visible: the clips that took in fabric at the back show on the back view.
- Product cropped to fit the ratio: a long coat squeezed into a square frame with the hem cut off. Amazon asks for the entire product in the frame; export a portrait or padded version instead.
The last two also cause returns rather than rejections. A shopper who receives a different shade or a longer coat than the photo showed sends it back, and the cost of that return is far higher than the time a second color check takes.
AI ghost mannequin generators: what they invent and what to disclose
Several tools now produce a ghost mannequin image from one photo of a garment on a mannequin or a hanger. They save the second plate and the masking. The catch is the part they have never seen: the inside of the neck, the label and the back lining are generated, not photographed.
- Labels and linings: a generated neck label can show a wrong size, brand mark or care symbol. Zalando asks for the inner back view, and shoppers zoom in on it.
- Shape: a generator can smooth wrinkles, change the drape or shorten a hem. Amazon requires images that accurately represent the product's scale and color.
- Metadata: Google requires images created with generative AI to keep metadata that marks them as AI-generated, such as the IPTC DigitalSourceType tag. Do not strip it when you resize.
A safe middle path is to keep the real inner neck plate and let software do only the masking and cleanup. Whatever you use, compare every output with the physical garment before it goes live.
Naming, variants and alt text for ghost mannequin images
Ghost mannequin shoots produce many near-identical files: the same T-shirt in eight colors, front and back, plus plates. Without a naming pattern the wrong color ends up on the wrong variant, and that mismatch is exactly what shoppers notice. Name each export after the variant SKU, then the view and a sequence number:
- TEE-ORG-NAVY-M_front_01.jpg, TEE-ORG-NAVY-M_back_02.jpg, TEE-ORG-NAVY-M_detail_03.jpg
- Keep plates out of the export folder (TEE-ORG-NAVY_plate-neck.tif stays with the masters).
- Shoot one sample size per color and link the images to every size of that color.
- Give each color its own main image. Amazon requires a main image for every child SKU, in that child's color.
Alt text should describe the garment, not the technique: "Navy organic cotton crew neck T-shirt, front view" rather than "ghost mannequin image". The free image alt text generator helps for one-off images. And check coverage: in WISEPIM's import snapshot of about 50 catalogs, 36% of products had no image and the median product had one, so a three-view ghost mannequin set is already above where many catalogs start.
In WISEPIM, Quality Guard's Media Readiness pack checks that every product has an image, that the gallery has at least three views, that images have alt text and that every variant has its own image. The Color Attribute Matches Image pack compares the main image's dominant color with the color attribute, which catches a navy shirt filed as black.
Preparing finished ghost mannequin images for every channel
WISEPIM does not shoot or composite ghost mannequin images, and it will not remove a mannequin from a photo. Its Image studio job in Catalog Jobs works on the finished files: it can put photos on a white background, pad them to a square frame with an even margin, make lighter files at a long edge of 1200, 1500 or 2000 px, and write alt text where it is missing, in the languages you choose. It never upscales, never crops into the product and makes no generative edits.
Each edit is staged as a new version in the review queue, so the live photo stays until you accept it, unless the run's auto-accept threshold lets confident edits through. For Zalando, switch the square frame off, because Zalando wants portrait 1:1.44 images, and choose the 2000 px long edge or leave the lighter file off, since Zalando's best-practice size is 2600 px tall. Where only alt text is missing, the Image Alt Text enrichment fills empty fields and keeps existing ones:
Fashion teams that manage many colorways and sizes can read more on PIM for fashion brands.

