Abstract
Neural shading offers a new paradigm for real-time graphics, replacing hand-crafted algorithms with compact neural networks that can be trained to reproduce complex appearance. In this three-hour course, we introduce the core principles behind neural shading and guide attendees from foundational theory to practical implementation. We begin with an accessible overview of optimization and network training, then demonstrate how a small MLP can learn and reproduce texture data, highlighting the key techniques needed to make tiny networks perform well in practice. The course then explores modern hardware acceleration—focused on coop-vec and shows how to deploy neural shaders efficiently in C++. We conclude with real-world applications, including Neural Texture Compression and Neural Appearance, illustrating how neural shading is already reshaping production workflows.
Course Intro
