The quick answer
Pick 100 Creatives if you sell clothing through your own store and paid social, your samples are already fit-approved, and you need on-model images for every colorway and size faster than a studio calendar allows.
- Apparel imagery for 100 SKUs starts at $10,000 per 100-SKU programme, quoted in writing as a fixed total before any work starts.
- Creative batches come back in 48 hours, and revisions within 48 hours.
- A brand intake before production records your model preferences, colorways and export specs, so the same face and the same olive hold across a drop.
- We make product and campaign images with Overs, an AI image tool also built by our founder (Overs and 100 Creatives are both brands of 2587689 Alberta Ltd.), and we publish our own render counts below.
Pick something else if your product sells on how it fits a real body in motion, such as denim, swim or shapewear, or your audience follows one model or the founder. Book a fashion photographer and a fit model for the first. Book a UGC creator agency for try-on content.
An earlier version of this page said 80 to 95 percent of apparel model photography could move to AI, that blind tests show shoppers can't tell the difference, that AI likenesses need no model release, and that brands save $50,000 to $250,000 a year in model fees. None of it had a source, and the release claim is wrong in New York. This version removes all four.
AI fashion models vs real models: three options, not two
A clothing brand asking about AI models is usually choosing among three things, and each one carries different paperwork.
A real model on a shoot. A person wears your sample and gets photographed. You pay a booking fee and a usage fee, plus the photographer, studio, styling and the samples shipped to set. The photo is proof of how the garment hangs.
An AI-generated model. A synthetic person who does not exist. The garment is rendered onto that person from your product photos and references. There is no booking and no usage term to renew, and also no physical proof of fit.
A digital replica of a real model. AI images of a person you have photographed before, made from their photos. You keep a face your customers know, and you take on a consent agreement for the replica itself, separate from the original shoot contract.
Fit is the part AI can't check for you
A real model wears the sample. If the size 30 trouser breaks two inches long, the photo shows it and someone on set pins it. An AI model never touched the garment. It draws a plausible version, and clothing goes wrong in plausible ways: a hem that lands at the wrong point, a pocket that shifts between the front and three-quarter view, a rib knit that renders smooth.
Our own records put a number on the effort. Before Overs existed, we built four campaigns by hand in 2026 with AI image models. Men's trousers on a model took 40 renders to get 11 usable images. Surf apparel on one consistent model took 163 renders for 12 keepers, because holding the same face and body across a set is harder than getting one good frame. Gummy supplement ads, with no person and no fabric, took 38 renders for 12. The set that needed the same person in every frame cost the most attempts per usable image of the four.
So the working rule is to confirm fit on a person or a fit model first, then use AI for the variations of a style whose fit you have already seen. Every generated image gets checked against the physical sample for length, rise, drape and color before it ships. Our guide to reviewing AI creative from an agency has the checklist, and when you need a real photo shoot covers the styles that should never skip the studio. Whether a generated image is allowed at all depends on the channel; AI fashion photography vs a traditional shoot goes through the Amazon, Google Shopping and ad disclosure rules.
Do you need a model's consent to make AI images of them?
In New York, yes, and in writing. The Fashion Workers Act took effect on June 19, 2025. It requires the client, meaning the brand, to get "clear and conspicuous prior written consent" for any creation or use of a model's digital replica. The consent has to state the scope, purpose, rate of pay and duration of the use. Civil penalties run up to $3,000 for a first violation and $5,000 for later ones, and models can sue for damages on top (Morgan Lewis summary of the Act, checked October 2026).
Model management companies face the same rule and must get that consent separately from the representation agreement. An old shoot contract granting usage "in all media" is the wrong document to lean on for a replica. Write a new one that names the replica.
H&M's 2025 announcement shows what a cleaner deal looks like. It said it would create 30 digital twins of real models, that the models would own the rights to their twins, and that they could be paid each time a twin is used, as on any campaign (ABC News report, April 2025, checked October 2026).
A fully synthetic model that resembles no real person sits outside the replica rule. Check that it really resembles no one: if a generated face lands close to a known model or creator, you have a likeness problem and no contract to fall back on. This section is general information, not legal advice. Have a lawyer in your selling markets read your model contracts.
Do you have to label AI-generated models in ads?
In the EU, from August 2, 2026, Article 50 of the AI Act requires disclosure of realistic AI-generated or AI-manipulated depictions. According to Davis+Gilbert's summary of the European Commission's guidance, that covers human avatars that do not resemble any identifiable person, and an AI-generated person in an advertising context is one of the guidance's own examples. The test is whether the image could pass as authentic when it is synthetic (Davis+Gilbert, checked October 2026).
If you sell into Europe, a photoreal AI model on a product page or in a Meta ad needs a disclosure. Decide where it goes, and how it looks on a 4:5 feed ad and a 9:16 story, while the campaign is still a brief. Retrofitting a label onto 200 finished images is slower than designing a space for it. The rules on static ad sizes for Meta tell you how much room each placement leaves.
Are AI fashion models cheaper than real models?
Per image, usually. But compare the right lines. A real-model shoot bill has a booking fee, a usage fee that grows with the term, territory and media you buy, plus photographer, studio, styling, hair and makeup, and samples. Ask your model agency for a rate card with usage spelled out. The usage line is often what surprises a brand that moves its product page shots into paid social, because a product page license and a paid media license are priced differently.
An AI-generated model has no booking fee and no usage fee. A digital replica of a real model brings talent pay back, because the consent has to state a rate. What you pay with AI instead is production attempts and review time: the 40 renders for 11 trouser images above, and a person checking every one against the sample.
We don't publish a per-image comparison, because it depends on how many styles, colorways and angles a drop needs. For reference, 100 Creatives quotes apparel imagery from $10,000 per 100-SKU programme and product photography projects from $5,000, each as a written fixed total before work starts, with nothing charged until you approve it. Pricing lists the starting points.
AI models, digital replicas and real models compared
| Real model shoot | AI-generated model | Digital replica of a real model | |
|---|---|---|---|
| Shows how the garment fits | Yes, the sample is on a body | No, it is rendered; check every image against the sample | No, it is rendered |
| Same face across a drop | Only if the model is rebooked | Possible, at a cost in attempts (163 renders for 12 keepers in our surf set) | Yes, that is the point |
| Consent paperwork | Shoot contract with usage terms | None for a person; confirm the face resembles no one real | Separate written consent naming scope, purpose, pay and duration (New York) |
| Talent fees that recur | Usage renewals | None | Pay per use, as the consent states |
| EU disclosure from August 2026 | Not for an unaltered photo | Required for realistic depictions | Required |
| Best for | Fit, try-on, hero campaigns, a known face | Colorway and size variants, ad variations | A known face across many SKUs |
More sizes and body types: what Levi's learned
In March 2023 Levi's announced a test with Lalaland.ai to use AI-generated models to supplement human ones and show more diverse models. Six days later, after criticism, it said it was not scaling back live photo shoots, live models or its commitment to working with diverse models, and described the test as a business efficiency (Rangefinder, checked October 2026).
For a smaller brand the lesson is narrow and practical. AI models can show a style on more body types than a one-day shoot can book, which helps a shopper picture it. They can't show how your size 18 fits a size 18 body, because nobody in that size wore it. If size range is part of your brand, shoot the extended sizes on real people first and keep generated bodies to variations of styles already shot in that range.
Which should you buy?
You're launching a new fit or a new block. A real model, or at least a fit model, before anything is generated. The first images of a new fit are the ones customers return items over.
You have 30 fit-approved styles in four colorways and a launch email in two weeks. AI models for the colorway and angle variants, checked against samples. That is 120 colorway images before ads, the volume where a studio calendar slips. The DTC clothing photography playbook has the shot list per style.
Your customers follow one model or the founder. Real shoots, or a digital replica under a written consent that names scope, purpose, pay and duration.
Your best ads are people trying clothes on. A UGC creator agency. AI models don't replace a creator's voice; see UGC ad creatives for how those ads get made.
You have an in-house designer and want to make variants yourself. An AI image tool. Overs is one, built by our founder, so weigh that disclosure; our comparison of an agency or an AI tool for product photos covers when each makes sense.
You sell into the EU. Whatever you pick for synthetic images, plan a disclosure for realistic AI people from August 2026.
Where 100 Creatives fits
100 Creatives is an AI-powered creative agency in Calgary, Canada, that makes static and video ads, and the product imagery behind them, for DTC and consumer brands advertising on Meta, TikTok and Google. For a clothing brand that means on-model product images for fit-approved styles and the apparel ad creatives built from them. Most of our client work is under NDA, so the portfolio shows spec work: concepts made for brands that did not commission them.
Send your style list, which styles are fit-approved, your colorways, size guide and launch date. If a style hasn't been on a body yet, we will tell you to shoot it first.