TL;DR. Running one artificial intelligence photoshoot is easy. Running it across two hundred SKUs takes two decisions made up front: who is in the frame and on what authority, and how frame two hundred stays matched to frame one. Both now have clear answers. The likeness rules are written down, New York’s synthetic performer disclosure law has been in force since June 9, 2026, and consistency comes from locking settings at the collection level rather than per image. This article covers both.
An artificial intelligence photoshoot compresses what used to be a studio booking into a browser session. Upload a reference photo, describe a scene, and a finished frame arrives in minutes for under $15, against $500 to $2,000 for a studio session. At that price a two hundred SKU catalog stops being a budget conversation and becomes a workflow one.
Generating one convincing image is solved. Every vendor will show you a single beautiful result. The work that remains is the part tool marketing never covers: deciding who appears in your images and on what authority, then keeping two hundred generated frames consistent enough to sit on the same category page. Both are workflow decisions with known answers, and this article walks through them. Marketplace and disclosure rules live in our existing coverage, linked below, rather than repeated here.
What actually happens in an artificial intelligence photoshoot
Quickly, because this part is well covered elsewhere. You upload a reference photo. The system segments the product from its background, which sets the ceiling on how real everything downstream looks. You describe a scene, or supply a reference image so lighting and color match a previous campaign. The model composes a new environment around the product with matched perspective, shadow direction and reflection.
If a person appears in the frame, you either use a model the platform generated or supply your own. That single branch carries more consequence than every other setting combined.
Who is in the frame
Fashion brands are working with three distinct layers of AI person imagery, and they carry different consent requirements.
Fully synthetic models
The system generates a person who corresponds to no real individual. No release exists to obtain, because nobody is depicted. The residual risk is accidental resemblance, since a generated face can land close enough to a real person to draw a right of publicity claim, and there is no release available for a face that was never photographed.
Generation from a real reference photograph
You start with your own shoot and generate variations. The person is identifiable, so the release governs, and the question becomes whether a release drafted before generative tools existed covers synthesizing new poses and settings from approved source frames.
This is the layer being tested in court right now. In June 2026, model Francheska Pujols filed suit in New York state court against Rainbow USA, claiming the retailer used AI to generate images of her in poses, settings and compositions she never posed for or approved, beyond the September 2024 release she signed. Rainbow’s response is that the contract already granted the right to alter, transform and composite her likeness for advertising. The court has not ruled. None of this means avoiding the layer. Reading the release before the generation run turns a court question into a twenty minute review.
Digital replicas of a specific model
A trained likeness reused across campaigns. This carries the heaviest consent requirements and the least ambiguity, because the law has largely caught up here.
Two New York statutes now sit directly on this. The Fashion Workers Act requires clear written consent, obtained in advance, before creating or using a model’s digital replica, and that consent has to specify scope, purpose, duration and compensation. Routine retouching is excluded. The Synthetic Performer Disclosure Law, in force since June 9, 2026, requires a conspicuous disclosure when an advertisement features a digitally created human who is not a recognizable real performer. Note the inversion between the two. One governs using a real person, the other governs using a fake one, and an apparel catalog with AI models can touch both.
The EU draws its line in the same place. Article 3(60) of the AI Act defines a deepfake as generated or manipulated content resembling existing persons, objects, places or events that would falsely appear authentic, and the Article 50(4) deployer disclosure attaches to that. A folded shirt on a generated background does not trigger it. A generated face that reads as a real, plausible person does.
Vision supports both paths. You can work with generated models, or upload custom photos of a real person where you hold the license, which is why the release question lands on you rather than on the tool. The fashion use case covers the apparel workflow in detail.
The practical rule: before a campaign ships, one named person should be able to say which of the three layers each image belongs to and produce the paperwork for it. That is a twenty minute conversation, and once it has happened the whole campaign ships with the likeness question settled.
How to keep an artificial intelligence photoshoot consistent
Consistency is what separates a demo from a catalog, and it slips in four specific places. Each one has a lock.
Identity drift. Run one prompt across forty SKUs and you get forty slightly different people. Different bone structure, different skin tone, different hair. Nobody notices on a single product page. On a collection grid showing twelve products at once, it reads as incoherent and quietly signals that the images were generated.
Lighting mismatch. Each generation picks its own key light. Hard afternoon sun from the left on one shot, soft overcast from above on the next. Both look fine alone. Side by side they look like four different brands.
Fabric and detail loss. This one hits apparel hardest. Generative models smooth texture. Knit structure, weave direction, drape and printed logo text are all at risk. A customer who receives a garment that does not match the weave in the photo returns it, and a return costs more than the image ever saved.
Scale errors. The model does not know how large your product is. A side table rendered beside a sofa can arrive the size of a footstool. In furniture that is a direct driver of returns, because the customer bought a dimension they inferred from the picture.
The fixes are unglamorous and they are all about removing choices from the model rather than adding prompt cleverness. Lock one lighting description and one model identity per collection. Generate in batches rather than one image at a time, so the whole run shares settings. Put a human on final review for physical logic, scale and text.
At catalog scale this stops being a tool question and becomes a pipeline question. If generation runs from your PIM or a script rather than a browser tab, the locked settings apply themselves and consistency stops depending on whoever is at the keyboard that day. Any tool you evaluate for a catalog of this size should offer programmatic access for exactly that reason. That is the difference between a tool you use and a process you run, and it is where a two hundred SKU catalog either works or does not.
Two constraints you inherit
Neither of these is specific to shooting with models, and we have covered both at length already.
Marketplaces set hard image rules that a lifestyle scene will fail if you point it at the wrong slot. Amazon’s main image must be a pure white background at exactly RGB 255, 255, 255, which is where most AI output quietly fails, and generated lifestyle work belongs in slots two through nine. The detail, including the metadata traps, is in AI product photos for Amazon.
Disclosure rules now differ by market and several deadlines have already passed. The full jurisdiction by jurisdiction map is in AI product photography regulations in 2026. The short version for a shoot with models is that a generated person raises your disclosure exposure well above a product on white.
Products that do not need a person
Everything above assumes someone in frame, because that is what photoshoot implies. A large share of the value does not need one.
Furniture, homeware, automotive and most hard goods work either way. A sofa in an empty sunlit room communicates scale, material and style with no likeness question at all. Add a person reading on it and you gain warmth and human scale, at the cost of re-entering the consent conversation above.
The method is identical. Segment cleanly, describe the environment, lock lighting across the collection, review for physical logic. What changes is that the no-people version removes an entire category of legal exposure and one of the four consistency failures, since there is no face to hold constant. For a first catalog run that is often the sensible place to start. We covered staging one furniture SKU for different buyer demographics in Generational style and generative AI.
What to do now
Classify your model imagery into the three layers and write down who holds the paperwork. Synthetic, generated from your own reference shoot, or trained replica. Each has a different answer, and the answer needs an owner rather than a file location.
Read your existing model releases before your next generation run, not after. If a release predates generative tools, assume it is contested rather than settled. That is the entire substance of the Rainbow case.
Lock lighting and identity at the collection level rather than per image. Treat the style brief as a fixed input for the whole run. Consistency is cheaper to enforce upstream than to fix in review.
Where managed services start to matter
None of this is beyond an in-house team. Locking a style brief, batching generations, tracking which images used which model layer, and keeping release records are all things a competent digital merchandiser can do. They are doable by hand, and for a small catalog they should be.
They are also repetitive, and the work grows with every SKU while the interest of the work does not. Vision handles the pipeline side, with batch generation and consistency controls across a collection, plus API and MCP access for teams driving it from their own systems once an account is set up in the app. The Done For You service exists for teams who would rather hand over the loop entirely. Plans start at $9.99/month and the pricing page lists credit allowances per tier.
The math changes somewhere around the point where nobody on the team can name every SKU.
One thing to do this week
Pick one collection. Generate the same model across five SKUs with your current settings, put the five faces side by side, and decide whether a customer would read them as the same person. If they drift, lock one model identity for the collection and rerun the batch. That is the whole fix, and it costs an afternoon.
References
Likeness and model releases:
- Rainbow argues existing model releases cover AI-generated ads, The Fashion Law, July 2026
- New York Synthetic Performer Law, Manatt, June 2026
- How AI, digital doubles and new laws are rewriting fashion and beauty, Foley and Lardner, March 2026
- Contracts, consent and clones, Lexology, June 2026
- Likeness, model releases and the legal framework for AI fashion imagery, May 2026
EU AI Act:
- Article 50, Regulation (EU) 2024/1689
- Code of Practice on Transparency of AI-generated Content, European Commission
Snapshot as of August 2026. Rules are moving, check the primary source before any compliance decision. This is general information and not legal advice.
