img-gen-gui

A lightweight desktop GUI for batch image generation using the Gemini API. Enter a prompt, choose how many images to generate, and the app fires all requests in parallel and saves the results as PNGs.

Features

• Batch generation — generate 1–10 images from a single prompt in parallel
• Secure API key storage — key is saved to the macOS Keychain via keyring; never stored in plain text
• Configurable output folder — defaults to ~/Desktop/genai, browseable from the UI
• Live log — dark-themed scrollable log shows model responses and saved file paths as they arrive
• Auto-open on macOS — each saved image is opened in Preview automatically
• Update notifications — checks GitHub Releases on startup and logs a link if a newer version is available

Requirements

• Python 3.10+
• macOS (Keychain integration; the rest of the app is cross-platform)
• A Google AI Studio API key with access to gemini-3-pro-image

Installation

git clone https://github.com/krmaynard/img-gen-gui.git
cd img-gen-gui
bash setup.sh

setup.sh creates a .venv virtualenv, installs dependencies, and prints the launch command. Run it once; re-run after pulling updates.

tkinter ships with the standard library. If it is missing on your system, install it via your package manager (e.g., brew install python-tk on macOS).

Install as a command

The project is packaged with pyproject.toml, so you can install it into any environment and launch it by name:

pip install .
gemini-image-gen

This registers a gemini-image-gen GUI launcher on your PATH.

Build a standalone app

To produce a self-contained bundle that runs without a Python install (a double-clickable .app on macOS, an executable elsewhere), use PyInstaller with the included spec:

pip install pyinstaller
pyinstaller gemini-image-gen.spec

The result is written to dist/ — dist/Gemini Image Generator.app on macOS, or dist/gemini-image-gen on Windows/Linux.

Usage

source .venv/bin/activate
python generate_image_gui.py

1. Click Set key and paste your Google API key. It is stored in the Keychain and only needs to be entered once.
2. Edit the prompt in the text box.
3. Set the Images spinner to the number of images you want (1–10).
4. Optionally change the Output folder.
5. Click Generate. Progress is shown in the log; images open automatically when saved.

Updating

The app checks for new releases on startup and prints a link to the log if one is available.

To apply an update:

git pull
bash setup.sh

Configuration

Constant	Default	Description
__version__	1.0.0	Current version; bump when cutting a release
MODEL_NAME	gemini-3-pro-image	Gemini model used for generation
KEYRING_SERVICE	gemini-image-gen	Keychain service name
DEFAULT_PROMPT	language-learning flowchart	Pre-filled prompt shown on launch

All constants are at the top of generate_image_gui.py.

How it works

Each generation request calls client.models.generate_content with response_modalities=["TEXT", "IMAGE"]. The model returns interleaved text and image parts; text is printed to the log and image data is decoded from the inline bytes and saved as a PNG via Pillow. All requests for a given batch share a single ThreadPoolExecutor so they run concurrently. The update check runs in a separate daemon thread at startup so it never blocks the UI.

Dependencies

Package	Purpose
google-genai	Gemini API client
keyring	Secure credential storage
Pillow	Decoding and saving PNG images
tkinter	GUI (stdlib)