> ## Documentation Index
> Fetch the complete documentation index at: https://docs.holala.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Model Faces
> Pick a model for your model templates, keep the same face across a whole drop, and choose between generating a new face or swapping one onto an existing frame.
Model templates put your product on a person. Before you Generate, you choose **who** that person is — from the model library, filtered to match your product and your market. Pick once, reuse everywhere, and the same face carries across your whole collection.
Model selection only applies to **model templates** — Product-to-On-Model, Model Swap, Jewellery Image Generation, Kids Image Generation, Lingerie & Swimwear Generation, and the rest. Object-only templates like White Background or Product Placement don't ask for a model.
## Pick a model from the library
Open a model template in the STUDIO, upload your product, then choose your model. The model library is built in — no casting, no shoot. Narrow it down with three filters until the model fits your product and audience:
Filter to the gender you're merchandising for.
Match the market you're selling into.
For close-crop templates — a hand for jewellery, a wrist for a watch, feet for footwear.
Match the face to your **product category and target market**. Fine jewellery and a hand model; streetwear and a younger, casual model; a luxury market and a model your customers recognise as one of their own. The closer the model fits the buyer, the more the image sells.
## Generate a face vs. swap a face
Two templates deal with the face itself. They sound similar but do opposite jobs — pick by what you already have.
**Creates a brand-new, consistent face** that doesn't exist yet. Use it when you want a signature model for your brand — one you can bring back for every drop.
**Puts a chosen face onto an existing frame.** Use it when the pose, outfit and scene are already right and you only need to change who's wearing it.
Reach for **Model Face Generation**. It builds a fresh face from scratch — no real person, no likeness rights to clear — and the result stays consistent so you can reuse it. This is how you get a house model that's uniquely yours.
Reach for **Model Face Swap**. The composition is done; you just replace the face with one from the library. Everything else in the frame stays put.
Generation **invents** a face. Swap **replaces** a face on an image you already have. If you don't yet have a frame to swap onto, start with Generation.
## Reuse one face across a whole drop
The point of a consistent face is repetition. Lock in one model and run it through every product in a collection — the same person in every shot reads as one campaign, not a stack of unrelated images.
Generate a new signature face, or pick one from the library and note it.
Use that same model on every product in the drop — each new product image, one consistent face.
Everything auto-saves to your Library, so your house model is there to bring back for the next launch.
Brand consistency comes from **holding the model steady** while the product changes — not from re-picking a new face each time. One face, one campaign.
## Good inputs still matter
A model template is only as good as the product photo you feed it. The AI dresses your model in the product you upload — a clean, correctly-angled input gives a clean result.
Plain background, single product, centered and fully in frame
Sharp and high-resolution — no blur, no crop
See [Product Images](/input-uploads) for the full product-photo checklist before you generate.
## Where to go next
Every template and the one job it does — including all the model templates.
How to prep the product photo, reference image and model selection.