- SIGGRAPH 2025 presentation on stochastic tile-based lighting techniques implemented in HypeHype
- cover motivations for the technique and how it compares against other solutions
- detailed discussion of the implementation
- covers how it interactics with other effects such as shadows
- presents techniques for implementing spectral rendering in real-time applications using Monte Carlo integration
- describes importance sampling strategies for wavelength selection based on illuminant spectra and color matching functions
- demonstrates that spectral rendering introduces only modest overhead compared to RGB rendering
- explains the mathematical foundations of tangent space and its role in normal mapping
- details how UV coordinates define the orientation of the tangent frame on surfaces
- presents the construction of the TBN matrix and its application in transforming texture-space normals to surface space
- discusses techniques to hide or reduce visible tiling patterns in terrain shaders
- presents shader based solutions
- additionally discusess alternative art authoring solutions
- introductory video for a course on neural shading techniques
- first part of an 11 video part series exploring what neural shaders are, what hardware they provide access to
- introduces a new Vulkan Drivers tab in the Vulkan Configurator (vkconfig)
- provides a graphical interface to add additional drivers, force specific physical devices, and reorder device enumeration
- streamlines multi-GPU testing and debugging by eliminating manual JSON configuration and environment variable tweaks
- video tutorial covering descriptor indexing in Vulkan
- explains how to access shader resources dynamically without binding individual descriptors
- demonstrates the modern approach to managing textures and buffers in Vulkan applications
- demonstrates how to use Nsight Graphics GPU Trace to profile and optimize RTX Mega Geometry
- shows detailed view of the available performance counter
- provides practical workflow for analyzing shader performance and identifying bottlenecks
- tiny dependency-free C99 library (~700 LOC) for lossless compression of GPU-compressed texture blocks (BC1, BC3, BC4, BC5)
- exploits spatial patterns, endpoint deltas, and Morton-ordered indices specific to BC formats achieving compression ratios of 1.5x-6x
- features deterministic bit-exact reconstruction with no malloc or external dependencies
- deep dive into GPU performance details and why GPUs are effective for neural network training
- contains high level overview of gpu achitecture
- identifies common bottlenecks (memory bandwidth, kernel-launch overhead, CPU–GPU imbalance) and how to overcome them
- tutorial on implementing Graph Neural Networks from scratch
- provides step-by-step explanation of GNN architecture and components
- covers fundamental concepts and practical implementation details
Thanks to Graham Wihlidal for support of this series.
Would you like to see your name here too? Become a Patreon of this series.
