For models and modeling agencies
Test the look before the casting.
Tarzi does not replace test shoots or a model's book. It previews: how a dress sits on a model, which frames a client might want, what a look does in motion — in minutes, from a photo of the garment and, with her agreement, a photo of the model.


What you get
Previews for clients and bookers
Show a client the garment on a model before a casting or a fitting, with the same face across six frames and a short film.
The same model, the same face
Add a photo of the model as a reference and the frames keep her identity. Use it only with her consent.
Originals you control
Every file is yours to download or delete. Nothing is used publicly by Tarzi.
How it works
Start from the garment
A photo or a design image of the piece the client wants to see.
Add the model photo, with consent
A clear, front-facing photo works best. The garment image stays the authority for the clothes.
Shoot six frames
Full length, three-quarter, detail, back, movement, portrait — same model, same setting.
Share or discard
Download the originals to send, or delete them; deleted files are purged from storage within two weeks.
Examples from the studio

Three-quarter frame 
Back view, same model 
Detail frame at the waist
Every picture on this page is Tarzi output made for the showcase, not customer work. Your own results depend on your references and your description.
What to expect
These are previews. A likeness is close but not exact; hands, jewellery and fine features are the usual tells. Do not present generated frames as a model's real photographs.
Consent and rights: upload a person's photo only with their agreement, and keep generated images of them under the same rules you apply to real ones.
Each run is different; a generation that fails returns its credits. Downloads are the original files as generated.
Questions people ask
- Can an agency build a digital portfolio with this?
- As a preview layer, yes: show bookers how a model carries a particular garment or styling. A model's book still needs real photographs, and generated images should be labelled as such — the files carry an embedded AI-content marker.
- How close is the likeness?
- Close enough to recognise, not identical; expect differences in fine features and hands. A clear, front-facing photo gives the best result.
- Who owns the images?
- Ownership of the generated file sits with the account that made it, as the Terms of Service set out. The person pictured keeps her own rights, so use the images with her agreement.
- Is the work private?
- Yes. Images are private to your account, never used as examples or for marketing, and you can delete them at any time.
Related pages
Try it on a design of your own
Open the studio, start from a sentence, a sketch or a photo, and judge the result on your own work. Credits are bought once and spent per piece; nothing renews.