Image Preparer Getting Started

Getting Started

Install with the launcher (recommended)

1. Open the RulZ AI Tools Suite launcher and click RulZ Image Preparer (under AI Training).
2. Click Install and wait. The launcher:
   • installs Python first if your PC doesn't have it,
   • downloads the app,
   • makes a private Python environment for it (a .venv folder inside the app's folder), so it doesn't interfere with anything else on your PC,
   • installs the parts the app needs (PySide6, OpenCV and NumPy, about 250 MB).
3. Click Launch.

The AI parts are not installed at this point. You only need them for Best quality (AI) mode and Auto crop, and the app offers to set them up the first time you use them (see below, and Auto crop's setup).

RulZ Image Preparer is in launcher version 1.7.0 and later. If you don't see it, update the launcher first (see Settings and Launcher Updates).

Updates show up as an Update button on the app's card. The AI parts you set up are kept.

The app is installed in %LOCALAPPDATA%\Programs\RulZ AI Tools\RulZ Image Preparer unless you changed the launcher's install folder.

Run it from source (without the launcher)

For people comfortable with a command prompt. You need Python 3.10 or newer.

1. Get the app's files. The branch app/image-preparer holds just this app, so you can clone it on its own:

   git clone -b app/image-preparer --single-branch https://github.com/speedyrulz/RulZ-AI-Tools-Suite.git "RulZ Image Preparer"

   (The same files are in apps/RulZ-Image-Preparer in the main repository.)
2. Double-click Launch RulZ Image Preparer.bat. The first time, it makes the .venv folder and installs the parts the app needs, then starts the app. After that it starts straight away.

Or do the same by hand, in the app's folder:

python -m venv .venv
.venv\Scripts\python -m pip install -r requirements.txt
.venv\Scripts\python run.py

To install the AI parts by hand too, instead of from inside the app, also run .venv\Scripts\python -m pip install -r requirements-ai.txt. The model file still downloads from inside the app.

First run

1. Click Add images and pick some pictures, or drag pictures or whole folders onto the window. Folders are searched all the way down, including subfolders.
2. Move the three sliders at the top. The thumbnails update as you go (they only show color noise removal; see why).
3. Click a picture to see it full size, before and after.
4. Click Save all… and pick a folder.

On the very first start, all three sliders are at 0, so nothing is changed until you choose a setting. After that the app remembers your slider values.

Setting up Best quality (AI) mode

The first time you click Best quality (AI) next to the JPEG slider, a window called Set up Best quality mode lists what it needs:

What	Size	Where it goes
PyTorch with NVIDIA GPU support (CUDA 13 build)	about 3 GB to download, more once unpacked	The app's own .venv folder
The FBCNN JPEG model (fbcnn_color.pth) from github.com/jiaxi-jiang/FBCNN	288 MB	%LOCALAPPDATA%\RulZ\Image Preparer\models

Click Set up. Installing PyTorch takes several minutes; you can watch its messages in the window. The model download then shows a progress bar. When it's done, the app checks the model file's size and checksum, so a broken download is never used.

If something fails, the button changes to Try again. Cancel stops the setup and keeps Balanced mode. Setup only does the parts that are still missing, so trying again doesn't download PyTorch twice. If it keeps failing, see Troubleshooting.

After setup, the app loads the model in the background so your first AI preview doesn't wait for it.