Clothing photography is the image set that sells a garment online: a front on-model main image, a back, a detail, a styled shot, and a flat laydown. Amazon publishes that set as a named file convention and mandates a crop for each garment category. Most of the work is compliance on one image, and free creative choice on the rest.
Every guide in this category opens with lighting, backdrops and camera settings, then presents flat lay, ghost mannequin and on-model as three styles you pick between. That framing is wrong for anyone selling on a marketplace. The method is not a preference. For adult apparel on Amazon, two of those three are on the prohibited list for the image slot that matters most.
This guide is the shot list. What Amazon names each file, how it wants each garment category cropped, which shots carry hard rules and which are free, and where AI covers the work without getting a listing pulled.
Key Takeaways
Amazon publishes a six-shot set, not a suggestion. Front on-figure (MAIN), back on-figure (PT01), detail on-figure (PT02), styled on-figure (PT03), laydown off-figure (FL01), and flat detail (PT13). It asks for a minimum of five and a maximum of seven images per product (Amazon Seller Imaging Guide, Clothing and Accessories, retrieved August 2026). “All adult apparel should be imaged on-model.” That is Amazon’s wording. Kids, baby, accessories, multipacks and sets go off-model instead. Ghost mannequins are banned from the main image, listed under Main Image Requirements DON’Ts alongside visible mannequins, grey backgrounds and promotional text. The technique half the internet calls the apparel standard will cost you the main slot. The crop is set by where the garment falls on the body. Full length at or below the knee, top body between fingertips and knee, waist down from above the navel to below the feet. Only the main image is constrained. Variants may show a full face, use coloured backgrounds, include props, and crop tighter or wider. Five of your six shots are creatively free. Resolution floor is 1,600 px on the height, with a 10,000 px ceiling on the longest side, at 72 dpi. Google requires AI-generated images to keep their IPTC metadata. Strip the DigitalSourceType tag and the items get disapproved in Shopping (support.google.com/merchants, accessed August 2026). Failure to meet the main image rules “may result in the removal of your ASIN from search.” Amazon states the penalty on the same page as the rules.
What is clothing photography?
Clothing photography is any image made to sell a garment: on a model, on an invisible mannequin, or laid flat. It differs from general product photography in one way that changes everything downstream. A garment has no fixed shape until something wears it, so the images have to prove fit, drape and fabric weight, not only colour and form.
That is why shoppers lean on them so hard. Apparel carries one of the highest return rates in ecommerce, commonly reported in the 20% to 40% band depending on subcategory, and fit is consistently named the leading reason. An image set that answers fit questions before checkout is doing returns work, not just marketing work.
The practical definition is narrower than the creative one. For a brand selling on a marketplace, clothing photography is a compliance task wrapped around a creative one. One image has to clear a published rule sheet. The rest exist to sell.
The shot list Amazon already wrote for you
Most brands build a shot list from scratch every season. Amazon has published one, with file names, and it maps cleanly onto any channel you sell through.
Each listing has a single Main image used in search and browse results, plus alternates called variants. Amazon names them with a variant code in the filename, following the pattern of product identifier, then variant code, then extension.
| Shot | Code | On or off figure | What it proves |
|---|---|---|---|
| Front | MAIN | On-figure | The garment, on a body, on white |
| Back | PT01 | On-figure | Rear construction and fit |
| Detail | PT02 | On-figure | Fabric, stitching, hardware |
| Styled | PT03 | On-figure | Context and brand voice |
| Laydown | FL01 | Off-figure | Flat shape and full colourway |
| Flat detail | PT13 | Off-figure | Texture at close range |
Amazon asks for “a minimum of 5 and maximum 7 images” per listing, citing customer studies, and says these should include a clear front and back image plus additional angles. It recommends adding a styled shot, multiple models with various fits, a close-up of details, and a video showing fabric and movement. One of the variant slots can hold that video.
The file naming matters more than it looks. The main image takes the MAIN code. Alternates take PT01, PT02, PT03 and upwards. For anything shot on-figure, Amazon recommends producing an off-figure alternate named FL01. Get the codes right and the images map to the correct slots on upload.
Six shots per SKU, five to seven slots, one of them constrained. That is the whole job.
How to frame each garment category
This is the rule almost nobody publishes, and it is the one that decides how you brief a shoot. Amazon prescribes the crop by where the garment sits on the body, not by product type in the abstract.
| Crop | Garment falls | Categories |
|---|---|---|
| Full length | At or below the knee | Dresses, suits and tuxedos, sleep sets, one-piece garments, robes and nightgowns, sweaters, jackets and outerwear below the knee |
| Top body | Below fingertips, above the knee | Tops and tees, one-piece swimsuits and rash guards, sweaters, jackets and outerwear above the knee |
| Waist down | From slightly above the navel or below the ribcage, down to below the feet | Pants and jeans, shorts, skirts, leggings and pyjama bottoms |
Read the outerwear rows together and the logic clicks. The same jacket gets a different crop depending on its length. A parka that falls below the knee is a full-length shot. A bomber is a top-body shot. The garment’s hem sets the frame.
Two consequences follow for a whole line. First, you cannot shoot every SKU at one focal length and crop later without losing resolution on the tighter ones. Second, a mixed drop needs its shot list sorted by hem length before anyone books a studio day, because that determines how many setups you actually need.
Athletic categories add their own layer on top of this, and the activewear shot list covers the crop rules that apply specifically to leggings and sports bras. Sweaters appear in two crop rows above because the hem decides the frame, but they also carry a capture problem no other category has, which knitwear photography works through gauge by gauge.
Which shots carry rules, and which are free
Here is the framework that saves the most time, and it comes straight from Amazon’s own comparison of mains against variants.
The main image must have a white background. Variants may use a different background. The main image is shown from the nose down. Variants may show a full face. The main image must have the product filling at least 85% of the frame. Variants may be zoomed further in or out. The main image must not have props. Variants may be fully styled with props included.
So the constraint is not distributed across your shoot. It is concentrated in one file.
That inverts how most teams budget a clothing shoot. The instinct is to spread effort evenly across six shots. The better allocation is to treat the main image as a technical deliverable with a checklist, then spend the creative budget on PT03, the styled shot, where nothing is restricted and brand voice is the only brief.
It also changes what you can safely generate. The five unconstrained shots have wide tolerance. The one constrained shot has a published rule sheet, and it is the shot to get right first.
What the main image must never contain
Amazon lists these as Main Image Requirements DON’Ts. The main image must not contain:
- Grey backgrounds
- Food, drink, handbags, accessories or props not included with the purchase
- Child or baby models, since all kids and baby items must be off-model
- Models kneeling, leaning or lying down
- Multiple product views in a single image
- Visible mannequins
- Ghost mannequins
- Promotional text such as “sale” or “free shipping”
- Logos, text, graphics or other badging that is not part of the product
- Packaging, brands or swing tags
The ghost mannequin line is the one that catches experienced teams. It is a widely recommended apparel technique and it is prohibited in this specific slot, while remaining perfectly legitimate in the alternates. We cover the full picture of where a ghost mannequin shot is allowed to run, including the channels and regions that take the opposite position.
The positive requirements sit on the same page. The main image needs a pure white background at RGB 255, 255, 255. It must show a single product, not stacked multiples, unless the listing is a multipack or set. The product must fill at least 85% of the image area. Visible hair should fall behind the shoulder and not cover any part of the product. Amazon accepts JPEG and TIFF here, and prefers JPEG.
On resolution, clothing images must be at least 1,600 pixels in height, must not exceed 10,000 pixels on the longest side, and should be 72 dpi. That 1,600 px floor is what switches on zoom, which Amazon says increases customer trust and sales.
Note the edition stamp before you treat any of this as final. The guide carries a Spring 2021 date and Amazon marks it as informational, so check the current style guide for your subcategory in Seller Central before a launch. Specs for the other channels you list on sit in our channel-by-channel image spec.
Where AI holds up on clothing, and where it breaks
Generated clothing imagery is genuinely good at some of these six shots and genuinely risky on others. The split follows the constraint map above.
Where it holds up. The styled shot, the second and third colourway of a garment you have already photographed, seasonal scene swaps, and the volume of crops and aspect ratios a paid social calendar eats. These are the unconstrained slots. Nothing is riding on a rule sheet, and the tolerance for interpretation is wide.
Where it breaks. Fabric behaviour under tension is the hardest thing to generate honestly. Knit stretch across a shoulder, the way a heavy twill holds a crease, how a bias-cut hem falls when someone moves. A generated garment that drapes like a lighter fabric than the real one is a fit claim you cannot support, and fit is the thing driving returns in the first place.
The second failure mode is consistency across a catalogue. A single generated shot can look excellent and still be unusable, because it does not sit beside the 200 images already on your site. That problem has a method rather than a fix, and we set it out in matching AI on-model shots to an existing catalog.
The honest position is that AI moves the alternates first and the main image last. Generate PT03 and the colourway variants. Photograph, or very carefully verify, the shot that carries the 85% rule and the white background.
DesignerBox works from your actual product photo rather than a text prompt, which is the difference that matters for apparel. Virtual try-on puts a real garment on a model, Fashion Factory produces the multi-shot set from one source image, and the Fashion OOTD workflow reruns the whole thing for the next drop. Try-on and AI video need the Premium plan or higher, and the commercial licence starts on Pro.
For the shots that stay flat, the same source photo drives a flat lay, which is the FL01 slot in Amazon’s naming. Socks are the exception Amazon wrote down by name, and sock photography covers the three prohibitions they are carved out of.
The metadata rule that gets AI clothing images disapproved
This one is missing from every clothing photography guide currently ranking, and it will disapprove a whole feed.
Google requires that all images created using generative AI carry metadata indicating they were AI-generated. The mechanism is the IPTC DigitalSourceType property, with values including TrainedAlgorithmicMedia, CompositeSynthetic and AlgorithmicMedia. Google’s guidance is explicit that you should not remove embedded metadata tags from images created with generative AI tools (support.google.com/merchants, accessed August 2026).
The trap is not the rule. It is the workflow. Export presets, resizing scripts and some CMS upload pipelines strip metadata by default as an optimisation. An image that was compliant when it left the generator arrives at Merchant Center stripped, and stripping it is treated as a policy violation rather than an oversight.
Two things to check before a feed goes live. First, that your export step preserves IPTC data. Second, that whatever sits between your asset library and your product feed does the same.
While you are in Merchant Center specs, note the other change on the calendar. Google is retiring the separate apparel minimum in favour of at least 500 x 500 pixels for every product, with enforcement beginning 31 January 2027, and recommends 1,500 x 1,500 or above for best performance across listing formats. The ceiling is 64 megapixels and 16 MB per file. Watermarks and promotional overlays are prohibited outright.
Disclosure obligations are a separate axis from metadata, and they moved during 2026. We track what each channel now asks for in marketplace rules for AI-generated product images.
Resale channels invert the whole compliance picture, because you are selling one specific garment in one specific condition rather than a repeatable SKU. Depop shop photos covers the grid problem on that side, and Poshmark listing photos covers shooting a closet at volume.
What a full apparel line costs to produce
Run the arithmetic on the shot list rather than on a day rate, because the shot list is what scales.
A 40-piece drop at six shots per SKU is 240 images. A traditional production covers that across multiple studio days, with model, styling, retouching and the reshoots that follow any colourway that arrives late. The cost lands per shoot day, and the second colourway of the same garment costs almost as much as the first.
Generated images invert the unit. In DesignerBox, generating or editing an image costs 5 credits. Basic is $15 a month for 500 credits, which is 100 images. The 240-image drop above sits inside the Pro plan’s 1,000 monthly credits, at $35 a month.
The number that matters more is the second colourway. Once the source garment is shot, additional colourways, crops and scene variants come off the same file rather than off a new setup. That is where the compounding happens, and it is the honest version of the savings claim. The full worksheet, including the four costs that do not go away, is in what cutting fashion photography costs really means.
Two limits worth stating plainly. Video is priced per second of output and is by far the most expensive operation on the platform, so budget it separately from stills. Team collaboration and shared brand kits sit on the Ultra plan at $200 a month. Current numbers are on the pricing page, and the apparel tooling sits in Photo Studio.
FAQ
How many photos do I need for a clothing listing?
Amazon recommends a minimum of five and a maximum of seven images per product, citing customer studies. The set should include a clear front and back image plus additional angles showing details. A short video can occupy one of the variant slots.
Can I use a ghost mannequin for a clothing main image?
Not on Amazon US for adult apparel. Ghost mannequins appear under Main Image Requirements DON’Ts alongside visible mannequins and grey backgrounds. They remain acceptable in the alternate slots, and other channels take different positions, so check each channel you list on.
What size should clothing photos be for ecommerce?
Amazon requires clothing images at least 1,600 pixels in height, not exceeding 10,000 pixels on the longest side, at 72 dpi. Google Merchant Center moves to a 500 x 500 minimum for all products from 31 January 2027 and recommends 1,500 x 1,500 or above.
Do I have to label AI-generated clothing photos?
For Google Shopping, yes. Images created with generative AI must carry IPTC DigitalSourceType metadata such as TrainedAlgorithmicMedia, and removing that metadata violates policy. Check that your export and upload pipeline preserves it, because many resizing tools strip metadata by default.
Can I photograph clothing on a hanger?
For a marketplace main image, no. Amazon states all adult apparel should be imaged on-model, and hangers fall outside that. On your own store the rule does not apply, though hanger shots tend to read as shapeless next to on-model or laydown alternatives.
Is flat lay good enough for clothing photography?
As an alternate, yes. Amazon recommends producing an off-figure laydown named FL01 for anything shot on-figure. As a main image for adult apparel it does not meet the on-model requirement, though kids, baby, accessories, multipacks and sets are imaged off-model instead.
What background should clothing photos use?
Pure white at RGB 255, 255, 255 for the main image, with grey backgrounds explicitly prohibited. Variant images may use a different background entirely, which is where colour, scene and brand styling belong.
Does the crop change by garment type?
Yes. Amazon sets the frame by where the garment falls on the body: full length at or below the knee, top body between fingertips and knee, waist down from above the navel to below the feet. The same jacket gets a different crop depending on its hem.
Sources
- Amazon Seller Imaging Guide, Clothing and Accessories, Spring 2021 edition, retrieved August 2026. Main image requirements, prohibited images, mains versus variants, image types and file naming, high resolution and zoom, and the full length, top body and waist down category pages.
- Google Merchant Center image requirements and AI-generated content policy, support.google.com/merchants, accessed August 2026.
- DesignerBox plan allocations and credit costs, product brief, August 2026.
Marketplace image rules verified against Amazon’s Seller Imaging Guide for Clothing and Accessories and Google Merchant Center documentation as of August 2026. Amazon’s guide carries a Spring 2021 edition stamp and is marked informational, so confirm the current style guide for your subcategory before a launch. Individual results vary.
