• comparison of combinations of six different sampling strategies for disk area light sources and 9 sample sequences
• testing is done for wholly and partially illuminated regions
• the best result could be achieved with the…

• a technique that allows the generation of infinite output space from a small example texture (such a grass, snow) in a pixel shader only
  • for example tillable terrain textures
  • cheap to evalua…

• discusses problems with synchronization between unrelated work when using a single command queue
• showcases how two command queues can achieve better overall utilization at the cost of longer latency

• how to use two Vulkan queues to enable optimal scheduling for compute-shader based post-processing
• discussion how fragment → compute can introduce bubbles on the hardware
• uses two software queues with semaphore synchronizatio…

• this in-depth document (26 pages) covers how real-world measurement is used to achieve correct lighting in call of duty
• explanation of camera concepts
• getting relative EV differences between light and shadowed area…

• shader authors can define variations of shaders with a number of systems
• each system can inject code/resources/constants into the shader
• a material allows specifications of which shaders belong together, which systems they use and allow…

• many optimization lessons learned for mobile
• how to deal with shader preloading
• discussion of error detection in automated image comparison tests
• increased game brightness to reduce the use of display backlights to save battery and reduce heat
• using …

• implements foveated rendering through usage of log-polar transformation
• gbuffer is written as normal, attributes transformed into sub-resolution textures using a log-polar transformation
• shading is done in log-polar space before the results are tr…

• experience of using vulkan with AMD drivers (stock and open source)
• comparison of register usage with different shader compilers
  

• how the stylized vfx in Rime was implemented
• short overview of fire and smoke effect
• in-depth overview of the water rendering implementation

• summary of
  • cloud modeling and authoring system
  • lighting model
• high level overview of the implementation

• environmen…

• April 30th, Vulkan Developer Day in Montreal, hosted by Ubisoft

• UVs map into a gradient texture
• to change the c…

Exploring scriptable render pipelines in unity 2018.1 [wayback-archive]

• overview of the scriptable render pipeline
• walkthrough of the development of a pipeline for mobile VR, code on github
• lighting, fog, lightmaps, shadows, light/reflection probes, transparency

SymPy case studies, part 2: deriv…

Breaking down barriers – part 2: Synchronizing GPU threads [wayback-archive]

• explanations of barrier behaviour on a simplified GPU model
• how split barriers help to increase GPU utilization

Color: From Hexcodes to Eyeballs [wayback-archive]

• in-depth introduction into light theory, human perceptio…

Triangle Visibility Buffer [wayback-archive]

• in-depth walkthrough of the GPU triangle culling pipeline
• render into a single render RGBA8 render target, storing drawID + triangle ID
• memory usage comparison with deferred shading
• at shading time
  • lookup the triangle ID from the render target
  • interpo…

This week I am at GDC, therefore there will be no newsletter at the end of the week. If you would like to meet during the week, send me an email or message me on twitter. DMs are open.

The series will return on April 2nd.

Meta loader for Vulkan API

• dynamically loads vulkan entry points, bypassi…

Porting GPU driven occlusion culling to bgfx [wayback-archive]

• breakdown of how a gpu driven pipeline can be implemented with bgfx
• indirect buffer writing from the CPU not supported so using regular instancing instead

Breaking down barriers – part 1: what’s a barrier? [wayback-archive]

• look at d…

Rendering in Camera Space(ish) [wayback-archive]

• shows effects of floating point precision loss far away from the origin
• transforms rendering to be relative to the camera position in the world instead of the origin of the world
• this makes rendering precision independent from the position in the w…

Youtube - Wolfenstein 3D’s map renderer

• in-depth breakdown of how the Wolfenstein 3D renderer works
• explanation of the ray-casting logic
• derivation of the mathematics found in the code

Fast way to render lots of spheres [wayback-archive]

• spheres drawn in pixel shader using ray-sphere intersectio…

Descriptor Pool Challenges in Vulkan [wayback-archive]

• looking at the descriptor pool problem with multiple worker threads
• variable workload makes it difficult to correctly size the pools
• discussions of workarounds
• proposal of a flag for lock-free descriptor allocation
  • looking at the AMD driver …

The Peak-Performance Analysis Method for Optimizing Any GPU Workload [wayback-archive]

• how to use GPU counters to detect interactions between different hardware blocks
• discussion of case studies
• walkthrough of analysis, code examples to show optimizations and discussion of results

Pre-Multiplied …

Optimized Swapchain in Vulkan [wayback-archive]

• discussion of different approaches, overview of strengths and weaknesses
• recommends splitting command buffers into two groups
  • pre-acquire (everything that doesn’t write into the swapchain)
  • post-acquire (everything that writes into the swapchain)
• a…

Tightening Up the Graphics: Tools and Techniques for Debugging and Optimization

• aimed at beginners in graphics programming
• overview of common problems and causes
• renderdoc explanation
  • feature and interface overview
  • how to use it to detect problems
• GPU perf studio and how to profile your applica…

Experiments in GPU-based occlusion culling part 2: MultidrawIndirect and mesh lodding [wayback-archive]

• using DrawIndexedInstancedIndirect (on AMD and Nvidia with API extensions)
  • all vertices in large buffers with manual vertex fetch in the vertex shader
• how to integrate level of detail for mes…

The Rendering of Middle Earth: Shadow of Mordor [wayback-archive]

• gbuffer breakdown
• blood rendering as gbuffer modifier
• tessellation
  • uses a point cloud as input and a tessellation shader calculates the polygons
• SSAO has two distinct channels that get applied to specular and diffuse separately
• t…

Unreal Engine 4 Rendering Part 1: Introduction [wayback-archive]

• focus on the deferred shading pipeline
• settings required for good development experience
• how data is passed from updated thread -> engine thread -> gpu
• how shader system is structured
• binding between HLSLS and C++
• how the corr…

White Paper: Foveated Rendering [wayback-archive]

• overview of typical VR rendering pipeline
• lens distortion causes pixels towards the edges to be distorted more
• Foveated takes advantage of this and uses a variable quality across the screen
• how to implement it using the opengl multiview extension
…

PIX 1711.28 – GPU memory usage, TDR debugging, DXIL shader debugging, and child process GPU capture [wayback-archive]

• can track d3d12 heap usage during timing captures
• experimental TDR (Timeout Detection and Recovery) debugging support
• shader debugging for DXIL (new intermediate shader language)
…

Decima engine: visibility in horizon zero dawn [wayback-archive]

• seperate system for statics and dynamics
• world broken into tiles
• sort key is used to define clusters
  • lod ranges, filter masks
  • morton numbers for spatial partioning
• tile/cluster culling on the CPU
  • launch one GPU culling job for ea…

Last Week on DirectX Shader Compiler (2017-11-14) [wayback-archive]

• support for explicitly sized types 16 to 64 bits
• spir-v improvements

Embree v2.17.1

• Improved performance of occlusion ray packets by up to 50%

Experiments in GPU-based occlusion culling [wayback-archive]

• An GPU occlusion system…