Getting Started

Getting Started

This guide helps you get Radiance running locally.

Requirements

• NVIDIA GPU with CUDA support (recommended) or appropriate ROCm support
• Python 3.10+ (use pyenv or a virtual environment)
• git
• Optional: a CUDA/CuDNN compatible environment for best performance

Quick install (recommended)

1. Clone the main repo git clone https://github.com/FXTD-Studios/radiance.git cd radiance

2. Create and activate a virtual environment python -m venv .venv source .venv/bin/activate # macOS / Linux .venv\Scripts\activate # Windows (PowerShell/CMD)

3. Install Python requirements pip install -r requirements.txt

4. GPU drivers and frameworks

• For CUDA (NVIDIA): install drivers and CUDA toolkit version matching the project's tested environment. See the repo README for tested versions.
• For ROCm (AMD): see ROCm installation guides and verify compatibility.

Running a minimal example

• There are example scripts in the repository (see /examples). A minimal run: python run_inference.py --config configs/default.yaml --input examples/input.png --output out/result.png

Configuration

• Configuration files live in configs/. Copy configs/default.yaml and adjust model paths, device settings, and precision (fp16/fp32).

Troubleshooting

• If you encounter out-of-memory errors, lower the batch size or use fp16 where supported.
• Ensure correct CUDA/cuDNN versions and that your GPU drivers are up-to-date.