- the GDC talk presents a great explanation of Quaternions and expands the concepts to cover dual quaternions
- presents interactive and visual examples of the different components and operations involved
- shows issues and limitations with matrix interpolations and how quaternions and dual quaternions can solve these limitations
![](/img/posts/graphics-programming-weekly-338/gdc-quaternions.png)
- this blog post covers how to extend a raytracer in a voxel world with support for reflections
- expands perfect reflection to allow for refraction
- contains challenges for the reader to develop their understanding further
![](/img/posts/graphics-programming-weekly-338/voxel-reflections.png)
- the paper introduces a method for view synthesis of facial expression captures in 3D views not captured
- the presented solution is running at real-time rates without relying on machine learning techniques
- achieved using a layered mesh for view-dependent information and mixing this with view-independent RGBA texture video
- the included paper presents a summary of the technique, compares it against existing solutions, and discusses the limitations of the approach
![](/img/posts/graphics-programming-weekly-338/face-method.png)
- the video presents a summary of all the papers that will be presented at the I3D Conference 2024
- shows a brief summary of a couple of seconds for each paper
- papers cover an extensive range of topics, from ML techniques, VR research, upcoming display technologies, new filtering approaches, and much more
![](/img/posts/graphics-programming-weekly-338/i3d-paper-preview.png)
- the paper discusses a generative model that is trained to generate multiple types of noise and blend between them
- presents the network architecture, how it has been implemented and improved
- shows many examples of the generated results
![](/img/posts/graphics-programming-weekly-338/noise_paper_teaser.jpg)
- the paper presents a framework to shape rendering noise to optimize samples for perceptual quality and denoising performance
- shows how sampling patterns can be optimized to take advantage of knowledge about post-processing spatial and temporal filtering characteristics
- demonstrates how to use the framework on example applications
![](/img/posts/graphics-programming-weekly-338/filtered-noise.png)
- the video shows the limitations of Nanite from an artist’s perspective
- presents the debug views to explain why vegetation scenes are more difficult for the system to process
- discusses possible solutions through the use of procedural generations and fallback LODs
![](/img/posts/graphics-programming-weekly-338/nanite_for_artists.png)
- the Godot engine is switching the definition of near/far plane, with the near plane now mapping to a depth of 1 and the far plane to 0
- this post explains common patterns that are affected by this change
- additionally provides links to explain why this new mapping allowed significantly improved depth buffer precision
![](/img/posts/graphics-programming-weekly-338/introducing-reverse-z.png)
Thanks to John Burton for support of this series.
Would you like to see your name here too? Become a Patreon of this series.