Licenses & Open-Source Attributions

Why Open Source Matters: Building on Giants

Media-Compute wouldn't exist without the open source community. The tools we use—from the React framework that powers our interface to the PyTorch library that runs our AI models—represent millions of hours of work by thousands of developers worldwide. We believe in acknowledging this work and contributing back where we can.

Open source software accelerates innovation by allowing developers to build on each other's work. Instead of every company reinventing basic functionality, we share common infrastructure and focus our efforts on creating unique value. This collaborative approach has produced much of the software that powers the modern internet.

Understanding License Types

Different open source licenses grant different rights and impose different obligations. The libraries we use fall into several categories:

AI Model Attributions

The AI models powering Media-Compute are built on research and pretrained weights from the machine learning community. While we've optimized and fine-tuned these models for our specific use cases, we acknowledge the foundational work of researchers worldwide.

Image upscaling capabilities are based on ESRGAN (Enhanced Super-Resolution Generative Adversarial Networks) architecture and its derivatives like RealESRGAN. These models were developed by researchers at various institutions and released for public use. Face enhancement features draw from GFPGAN (Generative Facial Prior GAN) research.

Image compression AI leverages research into learned image compression, including work published at conferences like ICLR and NeurIPS. We implement these techniques using PyTorch, which itself represents massive contributions from Facebook AI Research and the broader deep learning community.

Frontend Dependencies

Our web interface relies on a modern JavaScript ecosystem. Next.js provides server-side rendering and routing. React handles component-based UI development. Tailwind CSS enables rapid, consistent styling. These tools collectively save us thousands of hours of development time while providing a premium user experience.

Additional frontend dependencies include animation libraries, icon sets, and utility packages—all open source. When you experience smooth transitions, intuitive interactions, and responsive design, you're benefiting from the work of countless open source contributors.

Backend Infrastructure

Our Python backend uses FastAPI, a modern web framework that provides automatic API documentation, request validation, and high performance. FastAPI builds on Starlette for web functionality and Pydantic for data validation—both open source projects.

Media processing relies heavily on FFmpeg, a comprehensive multimedia framework that handles video encoding, format conversion, and audio processing. FFmpeg has been in development since 2000 and remains the backbone of countless media applications. We also use OpenCV for computer vision tasks and Pillow for image manipulation.

GPU and CUDA Dependencies

GPU acceleration is powered by NVIDIA's CUDA toolkit and cuDNN libraries. While not open source, NVIDIA provides these tools free for use. TensorRT enables optimization of neural network inference, dramatically improving processing speed. PyTorch's CUDA integration makes GPU programming accessible to Python developers.

Our Commitment to the Community

We believe in giving back to the community that makes our work possible. When we develop improvements or fixes for open source projects we use, we contribute them upstream where appropriate. We also support open source development through sponsorship and by hiring developers who contribute to community projects.

This license page serves as our formal acknowledgment of the projects we depend on. Each library listed here represents real work by real people, often done without compensation, for the benefit of the broader developer community. We're grateful for their contributions.

We strive to maintain accurate and complete attribution for all open source software we use. If you notice any missing or incorrect attributions, please contact us—we take license compliance seriously and will promptly address any issues.