- explains how to combine sampling of projected spherical caps and multiple importance sampling
- source code provided
![spherical_caps_importance_sampling.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/spherical_caps_importance_sampling.png&key=f3c6ead70c4785f58a9fc1deffe41401ac7f7b2664cbd4fb4b3a432ce5e48e21)
- collection of VFX tweets of the week
- tutorials, demos, and showcases of great looking effects
![technically_art_22.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/technically_art_22.png&key=dbfde1ed6e6edb29c67c3b7aa854161fdffc761ae010c365b787b6395cd9c51f)
- an updated article explaining color blending with OpenGL
- how the blend operators work and what premultiplied alpha is
![gl_blending.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/gl_blending.png&key=62d937590ddd73520e04860c32233f8d3faca49be1cc05af78b759157289002f)
- hardware overview of the new architecture for AMD GPUs
- now uses 32 or 64 waves wide operations
- unified cache hierarchy including compression for all stages
![rdna.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/rdna.png&key=f1409b42aeb5617ac2b5b9272f3f33fe77e735ac7b9c5f985ea24c1d6eefaa65)
- the article shows what changes have been introduced with Metal 3
- including metal support for the simulator
- cleaned up feature levels into families
![two_targets.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/two_targets.png&key=59d9c528abd13d8fe3ff37bcdd1dc857990500e7f7e99f0d8ebc7bc84ba44e44)
- the article shows what changes have been introduced with Metal 3
- covers additional details compared to the previous article
![metal_api.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/metal_api.png&key=d2f58b763605368d43b61d5fdbfa99cd1cf69bef47bd0be230445db4a70f4b1a)
- the paper investigates using the raytracing hardware support added in Turing GPUs to accelerate other tasks
- uses the hardware to detect in which tetrahedral mesh a point is located
- can achieve better performance than pure CUDA implementation
![rtx_beyond_rays.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/rtx_beyond_rays.png&key=aa8086a9a33796c227c5ec2590046d2ba73f0fb633c1ab47fd4c73be138453e4)
- breakdown of two frames in the game
- very little instancing
- all render targets are manually packed into 32-bit formats
- rats use a 4 level LOD system, and only the last level is instanced
![Final_frame1.png](https://www.jendrikillner.com/img/posts/graphics-programming-weekly-89/Final_frame1.png&key=62b36a464d11b8084f4bca98197dfff58694b3822cc1d4a948f65735acd29529)
Thanks to Jhon Adams for support of this series.
Would you like to see your name here too? Become a Patreon of this series.