- the article presents the flow of data when a CPU generated value is accessed from an HLSL shader
- explaining the differences between the available approaches as well as the implications on memory lifetime
- the article provides a summary of the Lumen implementation found in UE5 (Unreal Engine 5)
- discusses the offline and runtime stages of Lumen (dynamic GI solution)
- shows the Acceleration Structure tracing approach as well as probe placement
- presents how the scene is represented using MeshCard (Proxy) approximations and distance fields
- additionally shows a detailed walkthrough of all the logical stages that compose the system
- the article provides a summary of the features that are mandatory in Vulkan 1.3
- (removing requirement of render passe, better pipeline creation control, as well as direct buffer addressing support and more )
- additionally introduces the new Vulkan roadmap, this will provide guidance on what hardware/software support the group will focus on
- the 2022 roadmap will require bindless support as well as more consistent subgroup operation support
You are helping our core team to develop cutting-edge 3D data optimization technology, being used in production pipelines to process millions of 3D data sets each year, fully-automatically. You are performing research on 3D mesh processing, texture baking, UV mapping, optimization algorithms and ML-based algorithms for 3D data optimization and QA.
- the article explains the history and evolution of the mesa shader compiler (internal representation) NIR
- discusses the historical reasons and experience of using it instead of LLVM
- presents problems and advantages over a CPU focused IR system
- the new version of the Nsight Compute allows a better understanding of memory traffic and cache eviction decisions
- additionally adds new guided optimization support that detects unfused floating-point operations as well as uncoalesced memory access
- the article why randomly tracing rays leads to noisy results
- presents the mental model to improve this in a simple scene
- the presentation contains a collection of best practices when working with WebGPU
- covering compressed textures, debug label usage, data uploading, pipeline creation as well as bind group usage
- the video explains the derivation of Cosine Interpolation
- discusses how the individual components are derived and how they are connected for the final result
Thanks to Marius Horga for support of this series.
Would you like to see your name here too? Become a Patreon of this series.
