GIF Generator AI

 ╔═══════════════════════════════════════╗
 ║   ██████╗ ██╗███████╗  ██████╗ ███████╗███╗   ██╗   ║
 ║  ██╔════╝ ██║██╔════╝ ██╔════╝ ██╔════╝████╗  ██║   ║
 ║  ██║  ███╗██║█████╗   ██║  ███╗█████╗  ██╔██╗ ██║   ║
 ║  ██║   ██║██║██╔══╝   ██║   ██║██╔══╝  ██║╚██╗██║   ║
 ║  ╚██████╔╝██║██║      ╚██████╔╝███████╗██║ ╚████║   ║
 ║   ╚═════╝ ╚═╝╚═╝       ╚═════╝ ╚══════╝╚═╝  ╚═══╝   ║
 ╚═══════════════════════════════════════╝

A command-line tool to create GIFs from AI-generated images using Apple's Metal framework for GPU acceleration.

System Requirements

• Apple Silicon Mac (M1/M2/M3)
• macOS 12.0 or later
• Docker Desktop for Mac
• At least 8GB RAM (16GB recommended)

Installation

The recommended way to run this tool is using Docker:

# Build and start the container
docker-compose up -d --build

# Generate GIFs using the commands below (add 'docker-compose exec gif-generator' before each command)
docker-compose exec gif-generator python generate_ai_image.py --help

Usage

Creating a GIF using AI-generated images

Generate a GIF directly from a prompt:

python generate_ai_image.py "your prompt" -o output.gif -f 3 -s 20

Options:

• --output or -o: Output GIF filename (default: output/ai_image.gif)
• --steps or -s: Number of inference steps per frame (default: 50)
• --guidance or -g: Guidance scale (default: 7.5)
• --frames or -f: Number of frames to generate (default: 3)

Performance Notes

• The app uses Apple's Metal framework for GPU acceleration on Apple Silicon Macs
• NVIDIA CUDA is supported for systems with NVIDIA GPUs
• Image generation is significantly faster with GPU acceleration (Metal or CUDA)
• CPU fallback is available but much slower (approximately 10-20x slower)
• Recommended to use at least 20 steps for good quality images

GPU Acceleration Troubleshooting

For Apple Silicon Macs (M1/M2/M3):

• Ensure you're using macOS 12.0 or later
• Make sure PyTorch 2.0+ is installed (included in requirements.txt)
• The app will automatically detect and use Metal for acceleration
• If you encounter issues, try setting the environment variable: export PYTORCH_ENABLE_MPS_FALLBACK=1

For NVIDIA GPU Users:

• Install PyTorch with CUDA support: pip install torch==2.2.0 --extra-index-url https://download.pytorch.org/whl/cu118
• Ensure NVIDIA drivers are properly installed
• For Docker users, make sure to use NVIDIA Container Toolkit and add GPU runtime to docker-compose.yml