https://interjh.github.io TAA has some problem on UE Porjects History buffer is one of the reason. Anyway, I wanted to try running FSR 3 on my Mac before diving deep into TAA. FSR 3 is not currently supported on Mac. There is lots of problems. So i fixed the issue then check... Sat, 16 Aug 2025 00:00:00 -0700 https://interjh.github.io/2025-08-16-post/ https://interjh.github.io/2025-08-16-post/ It was simply out of curiosity. CitySample is not working with animation sharing. Mass Representation subsystem has special LOD system with ISM(Instanced Static Mesh) that can animate based on texture. This system is quite good to increase the performance. Still, High res geometry working with animation sequence. Implementation and application... Tue, 03 Jun 2025 00:00:00 -0700 https://interjh.github.io/2025-06-03-post/ https://interjh.github.io/2025-06-03-post/ Performance tests. You can easily find out which one is the nanite based on prims, draws. Performance tests Render Type Results Nanite Traditional So, As you can see the stat. Nanite one is quit slow on M1Max. But it is not correct compare because the number of vertex are different.... Sat, 10 May 2025 00:00:00 -0700 https://interjh.github.io/2025-05-10-post/ https://interjh.github.io/2025-05-10-post/ I tried the codes below RWTexture2D<uint> OutDBuffer32; RWTexture2D<uint> OutVBuffer32; . . . void WritePixel(...) { . . . uint OldDepthValue = 0; InterlockedMax(OutDBuffer[PixelPos], DepthInt, OldDepthValue); if(OldDepthValue < DepthInt) { OutVBuffer[PixelPos] = PixelValue; } } Still have the problems. Some pixels are flikering. Mon, 05 May 2025 00:00:00 -0700 https://interjh.github.io/2025-05-05-post/ https://interjh.github.io/2025-05-05-post/ Xcode not working fine in many area. like Intellisense, file indexing, etc… I have been testing the rider application on my Mac for a week. It is much better! Fri, 02 May 2025 00:00:00 -0700 https://interjh.github.io/2025-05-02-post/ https://interjh.github.io/2025-05-02-post/ My Macbook Pro has M1 Max. It dose not support atomic 64 in metal shader. Just try to split the 64bit buffer into two 32bit buffers. One is for depth, other onw is for visibility. Here is the problem. Depth buffer can use atomic max. and works fine. but visibility... Thu, 01 May 2025 00:00:00 -0700 https://interjh.github.io/2025-05-01-post/ https://interjh.github.io/2025-05-01-post/ Occlusion culling based on coverage buffer. Downscale depth target. Readback to CPU, Readback latency will be serveral frames( It will be depends on graphics command queue counts. ). Reprojection with current frame camera matrix(Bottom-left depth image above the image). Loop each object CPU Rasterization of AABB or OBB. Visibility Z... Mon, 20 May 2019 00:00:00 -0700 https://interjh.github.io/2019-05-20-post/ https://interjh.github.io/2019-05-20-post/ This is my first post, how exciting! It will be long journey! Sun, 19 May 2019 00:00:00 -0700 https://interjh.github.io/2019-05-19-post/ https://interjh.github.io/2019-05-19-post/