How can I change the body type or size of models in my product images using AI?

Quick Answer

Generate a person with the build you need instead of digitally reshaping a real person's body. In Nightjar, the Fashion Model you select sets who appears: their face, hair, skin tone, and apparent age. To set the body type, describe it in Custom Directions, such as “a fuller, curvy build.” Save each Fashion Model and build, with the same Pose and Photography Style, as its own Recipe, then apply those Recipes to each new Product. AI can create representative product imagery across broad body types, but it cannot verify a clothing size, body measurement, garment fit, or how the same item will fit an individual shopper.

What is the safest Nightjar workflow for showing different body types?

The safer workflow is to cast the person and describe the body separately, while keeping the product direction stable. In Nightjar, a Fashion Model is a reusable AI person that can wear, hold, or appear with a product across multiple images. Every Fashion Model, pre-built or custom, is saved as an identity card that shows only the face and hair, and Nightjar passes only that card and the model's name to generation. No size, measurement, or build setting travels with the selection, so the face card is the only physical cue the image model gets about the person. Your written direction is what sets the build: Nightjar carries it into the prompt for every image in the request, and without it, the image model decides the build.

1. Give Nightjar strong product evidence. Create a Product, Nightjar's reusable subject record, with clear front, back, and detail photos where available. These references help preserve the garment, but every output still needs review.
2. Choose the Fashion Model for the face. In Product Photography, choose With model under Shot > Subject, then open Model and pick a pre-built Fashion Model or one of your own. The picker can be filtered by Age (18–25, 25–35, 35–45, 45–55, or 55+) and Gender (female, male, or neutral). Those are casting metadata, not body measurements. The cards show faces, and a face carries build cues, so choose a person whose face, hair, and apparent age plausibly fit the build you will describe.
3. Write the build in Custom Directions. Custom Directions, the Create form's field for written refinements, is where model appearance belongs. Describe the body the camera should see in plain, respectful words, such as “fuller build,” “petite frame,” or “tall, athletic build.” Prefer visual words to a size label: “size 18” or “XL” means different things across brands and markets, and Nightjar has no setting for a person's size or measurements to translate it. The direction guides how the person is drawn; it is not a measurable body-size control.
4. Keep the comparison visually fair. A Pose is Nightjar's reusable body arrangement, while Camera Distance sets a close-up, medium, or full-body crop within the selected Pose's supported choices. Every Pose card shows the same neutral figure, and Nightjar copies its arrangement without its proportions, so one Pose works across builds. Nightjar calls its reusable lighting, camera, and mood direction a Photography Style. Keep that Style, Pose, Camera Distance, background, and output settings stable so different bodies receive the same product prominence rather than stereotyped poses or roles.
5. Save each body presentation as a Recipe. A Recipe saves the Fashion Model, Custom Directions, Pose, Camera Distance, Photography Style, background choice, and output settings, but not the Product. Save one Recipe per body presentation, apply each to the same Product, then repeat with the next Product. Applying a Recipe replaces whatever is in Custom Directions with the Recipe's own, so add product-specific notes after applying it and keep the build words.

That last step is why Nightjar suits a size-inclusive catalog. In a general image chat, the garment, the person, the build, and the shoot are briefed request by request. In Nightjar, the Product holds the garment evidence and each Recipe holds one person, build, and shoot setup, so the brand's range of bodies becomes a saved casting decision any Team member can apply to the next garment, not a sentence someone has to remember.

A custom Fashion Model is an identity tool, not a body tool. Its builder accepts one to five source images of the same person plus a name, age range, and gender, then creates a face-and-hair identity card like the pre-built ones. Only that card is used in generation, so the source person's build does not carry over, and the builder does not accept clothing size or body measurements. Describe the build in Custom Directions here too, and do not present custom sources as a way to reproduce a precise body shape or the same identity without variation. If the sources depict a real person, describe the build they actually have: showing an identifiable person with a body they do not have is the reshaping this workflow avoids. Obtain permission to use their likeness and follow Nightjar's Terms, which prohibit infringing privacy or publicity rights and impersonation.

Can AI Fashion Models show how a specific size will fit?

AI Fashion Models cannot establish how a specific size will fit. A generated image may depict plausible drape and coverage, but it is not evidence that a garment in size 14, XL, or any other label will sit that way on a shopper with particular measurements.

Keep the product page's size chart, garment measurements, fabric and stretch details, and real fit guidance separate from the generated image. Do not label an AI output as a verified fit example for a particular size unless it is backed by real fitting evidence.

What should I review before publishing an AI body-type variant?

Review each output against the source garment, the product information, and the build you asked for. Check silhouette, length, seams, closures, prints, logos, proportions, coverage, fabric behavior, and whether hands or skin intersect the product incorrectly. Nightjar's built-in visual review can retry some obvious product or Fashion Model failures, but it does not check whether the body matches your direction, and it does not certify size or fit.

Disclose that imagery is synthetic when a shopper could otherwise interpret it as documentary evidence of a real fitting, testimonial, or endorsement, and follow the rules for the channel and market where it appears. If a fit-sensitive detail cannot be validated, the garment changes materially across Generations, or the image supports a consequential sizing claim, use a real fitting and reshoot instead.