Notes from 'Nanite GPU Driven Materials'
My notes from "Nanite GPU Driven Materials" by Graham Wihlidal. Describes the changes in Nanite after Unreal Engine 5.0.
My notes from "Nanite GPU Driven Materials" by Graham Wihlidal. Describes the changes in Nanite after Unreal Engine 5.0.
Using software rasterization to render hair. Covers point projection, quad software rasterization, and segment-space calculations.
My thoughts on UE5's Nanite after I reimplemented it. Article is inside the project's git repo.
Math notes for 3D Gaussian Splatting renderer.
A guide how to create, style, and simulate strand-based hair using "com.unity.demoteam.hair".
A guide to implement retrieval-augmented generation (RAG) app that can infer user's questions from chat context.
A guide to using WebRTC for in-browser p2p connections. Explains STUN and TURN servers as well as ICE protocol.
Let's take a look at the graphic and compute pipelines in Vulkan. Discover how to use uniforms and push constants. Get to understand VkGraphicsPipelineCreateInfo and much more.
Explanation of memory allocation in Vulkan. How to use AMD's Vulkan Memory Allocator library. Exploring different parameters for VBuffer and VkImage.
A closer look at Vulkan synchronization objects, vkCmdPipelineBarrier(), execution vs memory dependencies, synchronization with swapchain, frames in flight, and many more.
Walkthrough of Vulkan initialization steps. Create VkInstance and VkSurfaceKHR, pick the GPU device, prepare the swapchain, and more.
Learn how to debug Vulkan using RenderDoc. Add labels to objects and passes. Debug GLSL code as if it was run on CPU.
Some parts of OpenGL API reach as far back as the year 1992 and it shows. Let's look at how to manage the basic OpenGL state in a more stateless way.
My notes containing neural network backpropagation equations. From chain rule to cost function, gradient descent and deltas. Complete with Convolutional Neural Networks as used for images.
Tips for implementing neural networks from scratch. From identifying use cases and data generation to implementation, testing, debugging and profiling. A separate section is dedicated to GPUs and unique problems for massively parallel devices.
Various tests to see how markdown formatter reacts.