MoneyPrinterTurbo enables fully automated short-form video creation from simple text prompts. By supplying a subject or keyword, the system generates complete video scripts, sources royalty-free footage, creates synchronizde subtitles, and mixes background audio into a production-ready HD video.

Core Capabilities

The application employs a modular Model-View-Controller design, offering both RESTful API endpoints and an intuitive web interface. Key functionalities include:

• Automated Script Generation: AI-powered narrative creation with optional manual override
• Flexible Aspect Ratios: Native support for 9:16 vertical (1080×1920) and 16:9 horizontal (1920×1080) formats
• Batch Processing: Generate multiple variations simultaneously for selection
• Granular Control: Configurable clip duration and transition timing
• Multilingual Support: Chinese and English content generation
• Audio Synthesis: Integration with multiple TTS providers
• Subtitle Customization: Adjustable typography, positioning, color schemes, and stroke effects
• Audio Management: Randomized or specified background tracks with volume control
• Asset Sourcing: High-definition, copyright-free video libraries
• LLM Flexibility: Compatible with OpenAI, Moonshot, Azure, Qwen, Google Gemini, Ollama, and GPT4Free

System Requirements

Minimum specifications:

• CPU: 4+ cores
• RAM: 8GB+
• GPU: Optional (CPU rendering supported)
• OS: Windows 10+ or macOS 11.0+

Avoid installation paths containing non-ASCII characters to prevent path resolution errors.

Deployment Methods

Containerized Deployment (Recommended)

Prerequisites: Docker Desktop installed and running. For Windows, ensure WSL2 backend is enabled.

# Clone repository
git clone https://github.com/harry0703/MoneyPrinterTurbo.git
cd MoneyPrinterTurbo

# Launch services
docker compose up --build

Access points:

• Web UI: http://localhost:8501
• API Documentation: http://localhost:8080/docs or http://localhost:8080/redoc

Manual Installation

Create an isolated Python environment:

git clone https://github.com/harry0703/MoneyPrinterTurbo.git
cd MoneyPrinterTurbo

# Using conda
conda create -n video-generator python=3.10
conda activate video-generator
pip install -r requirements.txt

ImageMagick Installation (required for subtitle rendering):

Windows:

1. Download static binary from ImageMagick archives
2. Install to default location (avoid custom paths with spaces)
3. Update config.toml with installation path

macOS:

brew install imagemagick

Linux (Ubuntu/Debian):

sudo apt-get update && sudo apt-get install imagemagick

Linux (RHEL/CentOS):

sudo yum install ImageMagick

Launching Services:

Web Interface:

# Windows
conda activate video-generator
.\webui.bat

# Unix/Linux/macOS
conda activate video-generator
chmod +x webui.sh && ./webui.sh

API Server:

python main.py

Configuration

Duplicate config.example.toml to config.toml and configure:

LLM Provider Setup:

• OpenAI: Set llm_provider = "openai" and provide valid API key
• Free Alternative: Use g4f (GPT4Free) for zero-cost access with reduced reliability
• Moonshot: Configure with API key from platform regitsration (includes free credits)
• Qwen: Alibaba Cloud integration available

Free GPT-3.5 Setup: Deploy local proxy container:

docker run -d -p 3040:3040 missuo/freegpt35

Then configure:

llm_provider = "openai"
openai_api_key = "sk-dummy-key"
openai_base_url = "http://localhost:3040/v1/"
openai_model_name = "gpt-3.5-turbo"

Media Processing

Voice Synthesis: Multiple TTS engines supported. Refer to voice catalog for available options.

Subtitle Generation: Two provider modes available:

• edge: Fast processing, low resource usage, variable accuracy
• whisper: Higher accuracy, requires ~3GB model download, computationally intensive

Configure via subtitle_provider in settings. Leave empty to disable subtitles.

Asset Management:

• Background music: Store files in resource/songs/
• Custom fonts: Place in resource/fonts/

Troubleshooting

AttributeError: 'str' object has no attribute 'choices' Indicates LLM response parsing failure. Typically network-related. Verify VPN connection or configure proxy settings in openai_base_url.

RuntimeError: No ffmpeg exe could be found Download FFmpeg static builds and specify path:

[app]
ffmpeg_path = "C:\\ffmpeg\\bin\\ffmpeg.exe"

Network Errors (Audio/Video Download) Connectivity issues with external services. Enable global VPN tunneling for regions with restricted access.

ImageMagick Policy Restrictions Modify /etc/ImageMagick-X/policy.xml entries containing pattern="@", changing rights="none" to rights="read|write".

OSError: [Errno 24] Too many open files Increase system limits:

ulimit -n 10240

PIL Attribute Error (ANTIALIAS) Pillow 10.x compatibility issue. Downgrade to stable version:

pip uninstall Pillow
pip install Pillow==9.5.0