1. Start with one clear photo
Photograph the garment from the front, on a hanger or laid flat, with nothing cut off. Add the back if it has a print, and a close-up for small details.
2. Choose the model
Pick a ready-made model, generate one that fits your brand, or use a photo of a real person — only with that person's consent. Keep the same model across a collection so your pages look like one shoot.
3. Check what must not change
Before you publish, compare the result with the garment: colour, print, buttons, pockets and logos. Correct a detail rather than accept a close guess.
4. Keep the AI label
AI-generated images carry a marker inside the file. Upload the file as delivered; resaving it in an editor can strip that marker.
Source photos that work, and what stays hidden
Three kinds of photo give the try-on something to read. On a hanger the fabric falls under its own weight, so length and drape are visible; a hanger narrower than the shoulders lets the shoulder line sag. Laid flat, the garment shows its whole outline at once. On a shop mannequin or a dress form it is already in shape, with the front closed and the hem at its real length.
What the photo hides cannot be copied. A sleeve folded behind the body or a pocket covered by a hand is not in the picture, and the tool is briefed to keep the garment as the photo shows it. The reverse side is known to the tool only when a photo of the back is added; a back print is then briefed to stay on the back. An upload that shows only the back, or a product in its packaging, can be flagged before anything is generated.
One model, one size, one background per collection
Consistency starts with the person. Virtual try-on offers ready-made models, models saved from Create model and a photo of a person who has given permission. Create model defines a person by gender, age range, ethnicity, body type and hair, and a saved model can be selected for every product. With no model selected, the choice of person is left to the tool.
Size, pose and background follow the same logic. Sizes run from XS to XXL and the brief fits the garment as the size chosen, so one size across a range keeps the images comparable. Ten poses are offered, and Change pose renders a delivered image in another, briefed to keep the same person, garment, background and light. The background is a studio ground, an outdoor or lifestyle setting, a location or an uploaded image.
Checking against the garment and keeping the AI label
With the garment itself in reach, judge the colour in daylight, count the buttons and pockets, read any print at full size and compare the collar, sleeve length and hem. A label sewn inside the neck is briefed to stay inside, so one on the outside is a fault. A single wrong detail can be marked in the delivered image and described; only that area and its immediate surroundings are changed.
The delivered file carries a machine-readable marker that declares it AI-generated, and an image showing a person generated by Herofab also carries the keyword for AI-generated people. Both sit in the metadata of the file. An editor, a screenshot or a chat app can save a new file without them, so the file is uploaded as delivered. AI image label checker reads a file and reports which markers it still carries.
Checklist
- Front photo, whole garment in frame
- Back and detail photos if needed
- The same model across the collection
- Colour, print and details checked
- File uploaded as delivered, marker intact
Updated 24 September 2026