Image matting, optimization, and removal require you to deploy Tucools yourself and then download models separately. For the client, download the installer for your system from the [release page](https://github.com/dingdangdog/tucools/releases), then install and open it. For Docker, after downloading the [project](https://github.com/dingdangdog/tucools), execute `docker compose -f deploy/docker-compose.deploy.yml up -d` in the project folder, then open http://127.0.0.1:8080 in your browser. Functions like compression and cropping do not require models.
## Download Models
[model-pack.zip](https://github.com/dingdangdog/tucools/releases/download/models/model-pack.zip) (approx. 900MB). The release page is [models](https://github.com/dingdangdog/tucools/releases/tag/models).
After decompression, you should directly see two folders: `models` and `bin`. If it's `model-pack/model-pack/models`, remove the extra layer.
## Where to Place the Models
After placing them, open image matting, optimization, or removal, and click "Recheck".
For the **client**, place them in a fixed directory, where `models` and `bin` are directly present:
| System | Directory |
| --- | --- |
| Windows | `%APPDATA%\tucools\model-pack`. Press Win+R, type `%APPDATA%\tucools` and press Enter. If `model-pack` doesn't exist, create it. |
| macOS | `~/Library/Application Support/tucools/model-pack` |
| Linux | `~/.local/share/tucools/model-pack` |
For **Docker**, place them in the project folder, alongside `deploy`, not inside `deploy`. The startup file is in `deploy`, and it will look for `model-pack` in the parent directory:
```text
tucools/
deploy/
docker-compose.deploy.yml
model-pack/
models/
bin/
```
