Thank You & Acknowledgements
Rice Upscaler is built upon state-of-the-art AI research and the generosity of the open-source community. We express our deepest gratitude to the following authors and foundational projects:
🌐 Project & Source
Rice Upscaler Official Website
Official homepage of Rice Upscaler project, distribution and release updates.
Rice Upscaler GitHub
Open-source repository, issue tracker, technical documentation and releases.
⚡ AI & Upscaling Engines
DAT (Dual Aggregation Transformer)
State-of-the-art Transformer architecture for Image Super-Resolution (CVPR), preserving textures and crisp fine details.
Real-ESRGAN
Practical algorithms for general image and anime illustration restoration, removing compression artifacts and blur.
Lanczos Resampling
High-quality windowed sinc mathematical resampling algorithm, running near-instantaneously without GPU requirements.
🛠️ Core Resources & Frameworks
Tencent ncnn
High-performance neural network inference computing framework optimized for cross-vendor Vulkan GPUs.
Microsoft DirectML
Hardware-accelerated DirectX 12 machine learning library providing high-performance inference on Windows.
PyTorch
Leading open-source deep learning framework used for executing and developing state-of-the-art AI architectures.
Pillow (PIL)
The friendly Python Imaging Library and cornerstone of Python image manipulation workflows.