Trying a garment before buying it
Your photos in the person card, the piece you want in the garment card: the render shows the cut on your silhouette. Useful on an expensive or hard-to-return item, less so on a basic tee.
One person card, one garment card: the AI generates the piece worn in a realistic photo, then animates it into a runway video. The tool for fashion shops that want worn shots without a shoot.
The real workflow: the person card and the garment card combine into a worn photo, which becomes a 5-second runway video.
At a glance
The worn shot is the photo that sells a garment, and the most expensive to produce. This workflow generates it from two Reference cards: the person (2-3 face photos) and the piece (packshot or flat-lay). The prompt @-mentions both, so the model receives real photos from both sides.
That is where the consistency comes from: the face stays the person card's face, the garment keeps its cut, fabric and details. Swap the piece and keep the model; swap the model and keep the piece: every combination is one generation, not a new shoot.
Seedream 5 Pro handles fabrics and drape finely; the runway video runs on Kling 2.5 Turbo. To build a permanent virtual model for your shop, follow building a consistent character.
These are the files of the workflow above, exactly as the models produced them, unretouched. The starting point is shown when there is one.
Starting point
Render
Online fashion shops and designers who want to show a piece worn by their own models or by the customer.
Two or three sharp photos of the person from different angles, and a photo of the garment flat or on a hanger, well lit.
Fill in the person card
2 or 3 face photos from different angles, plus a description (build, style). This card is what guarantees the likeness.
Add the garment
Packshot, flat-lay or worn photo: the garment card describes cut and fabric, and provides photos of the real piece.
Generate the worn shot
The prompt combines both cards: pose, backdrop and light are set in one sentence. Regenerate to vary the pose without losing the piece.
Animate into video
The worn photo becomes a 5-second runway walk: the person walks towards the camera, the fabric moves naturally.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
Your photos in the person card, the piece you want in the garment card: the render shows the cut on your silhouette. Useful on an expensive or hard-to-return item, less so on a basic tee.
Every reference worn by the same virtual model, with no shoot and no casting. Change the piece, keep the person: each combination is one generation at five credits.
The worn photo becomes a shot where the person walks towards the camera and the fabric moves, for twenty-six credits with Kling 2.5 Turbo. That is what shows drape, which no photo does.
Three person cards, one garment card: the piece is shown on three silhouettes. Buyers increasingly expect it, and a classic shoot makes it unaffordable.
The chain also works for what is worn without buttoning. Glasses remain the hard case: lens and reflection need a macro in the product card.
The back and sides are then invented. Front, back and a macro of the signature detail, exactly as for the ghost mannequin.
Stripes, checks and logos distort more often than plain fabric. Generate a close-up before publishing, the reflex that avoids returns.
Putting a garment on someone without their consent is an image-rights problem, even for an internal test. Your photos or a virtual model, nothing else.
Yes: put your photos in the person card and the piece you covet in the garment card. Same workflow for a shop or for yourself.
Cuts and fabrics come out very well; complex patterns and logos deserve a check. A close-up is a good reflex before publishing.
The worn photo costs a few cents, the runway video about 26 credits. Prices are shown node by node.
Yes: your photos in the person card, the piece in the garment card. It is the same workflow for a shop or for yourself.
Cuts and materials carry over very well, complex patterns and logos deserve a check. A close-up before publishing is a good reflex.
No, and that is the honest limit of the tool: it shows the look, not the exact fit on your body. It replaces a mental projection, not a fitting room.
Yes, by duplicating the Image node with a different framing: front, back, three-quarter. Both cards stay wired in, so neither the person nor the garment changes between angles.
Virtual try-on: have AI wear your garment
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.