Google Veo Studio

简体中文文档

A modern desktop GUI application for generating videos using Google's Veo 3.1 model via the Gemini API. Built with Python and PySide6.

Features

• GUI Interface: User-friendly desktop interface, no command line required.
• Veo 3.1 Support: Utilizes Google's latest video generation model.
• Customizable Parameters:
  • Prompt & Negative Prompt
  • Aspect Ratio (16:9, 9:16)
  • Person Generation Safety Settings
  • Seed Control for reproducible results
• Real-time Logging: Integrated console output within the GUI for monitoring progress and errors.
• Modular Design: Clean code structure separating GUI, logic, and configuration.

Prerequisites

• Python 3.10 or higher
• A Google Cloud Project with Vertex AI / Gemini API enabled
• A Google Cloud API Key

Installation

Method 1: Using uv (Recommended)

uv is a fast Python package installer and resolver. This project is configured to use Tsinghua University mirror for faster downloads in China.

1. Install uv (if not already installed):

   # On Windows (PowerShell)
   powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
   
   # On macOS/Linux
   curl -LsSf https://astral.sh/uv/install.sh | sh

2. Clone the repository (or download the source code):

   git clone https://github.com/sihuangtech/veo-studio.git
   cd veo-studio

3. Create virtual environment and install dependencies:

   # uv will automatically create a virtual environment and install dependencies
   uv sync

4. Activate the virtual environment:

   # On Windows
   .venv\Scripts\activate
   
   # On macOS/Linux
   source .venv/bin/activate

Method 2: Using pip (Traditional)

1. Clone the repository (or download the source code):

   git clone https://github.com/sihuangtech/veo-studio.git
   cd veo-studio

2. Create and activate a virtual environment:

   python3 -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

Configuration

1. Copy the example configuration file:

   cp .env.example .env

2. Open .env and replace your_api_key_here with your actual Google Cloud API Key.

   Note: The application will verify this file and alert you if it's missing or configured with placeholders.

3. Network Proxy Settings (Optional): If you are in a region where Google services are restricted (e.g., Chinese Mainland), you can configure a proxy by uncommenting and setting HTTPS_PROXY in your .env file:

   HTTPS_PROXY=http://127.0.0.1:7890

4. Custom API Endpoint (Optional): If you are using an API gateway / reverse proxy to access the Google GenAI API, you can override the SDK base URL in your .env file:

   GOOGLE_GENAI_BASE_URL=https://your-proxy.example.com

Video Model Selection

Google provides multiple Veo video generation models. You can now switch between them directly in the GUI.

1. Locate the Model Selection dropdown in the left panel.
2. Select your desired model.
3. The selection is automatically saved to config.json.

Model Name	Version	Description
veo-3.1-generate-preview	Veo 3	Released on Oct 15, 2025. Generates 720p or 1080p video at 24 or 30 fps.
veo-3.1-fast-generate-preview	Veo 3	Released on Oct 15, 2025. Generates 720p or 1080p video at 24 or 30 fps. Optimized for speed and rapid iteration.
veo-3.0-generate-001	Veo 3	Released in May 2025. Shut down in Nov 2025, replaced by Veo 3.1.
veo-3.0-fast-generate-001	Veo 3	Released in May 2025. Shut down in Nov 2025, replaced by Veo 3.1.
veo-2.0-generate-001	Veo 2	Released in Dec 2024. Generates 1080p video at 24 or 30 fps.

Notes:

• All models generate 8-second videos (API returns integer duration).
• Veo 3 series models support native audio generation and higher realism.
• Ensure your Google Cloud account is allowlisted for the corresponding preview models.
• For more official documentation and model details, please refer to: Google Gemini API Video Docs

Usage

Run the graphical user interface:

python3 run_gui.py

1. Select the desired model from the Model Selection dropdown.
2. Enter your video description in the Prompt box.
3. (Optional) Enter a Negative Prompt to specify what to avoid.
4. Adjust Aspect Ratio and Person Generation settings as needed.
5. (Optional) Check Use Seed and set a number for reproducible generation.
6. Click Generate Video.
7. Wait for the process to complete. The log panel on the right will show status updates.
8. Once finished, the video location will be displayed, and a success message will appear.
9. Generated videos are saved in the output directory within the project folder.

Project Structure

• run_gui.py: Launch script for the GUI application.
• main.py: Command-line interface entry point.
• app/: Source code directory.
  • gui.py: Main GUI window implementation.
  • veo_client.py: Core logic for interacting with the Google GenAI API.
  • config.py: Configuration management.
  • utils.py: Utility functions.
• config.json: Stores user preferences (e.g., selected model).
• config.example.json: Template for configuration.
• .env: Configuration file for API keys.
• pyproject.toml: Project metadata and dependencies (for uv).
• uv.toml: UV configuration with Tsinghua mirror settings.

Troubleshooting

• API Key Errors: Ensure your API key is valid and has access to the Veo model in Google Cloud Console.
• Quota Limits: Video generation models often have strict quota limits. Check your Google Cloud quota if generation fails repeatedly.
• Logs: Check the "Console Output" panel in the application for detailed error messages.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

Maintained by SK Studio.

• Website: www.skstudio.cn
• Email: contact@skstudio.cn
• QQ Group: Join Chat