fastMarchingCubes

• Fast implementation of Marching Cubes in Unity, using Burst and SIMD instructions.
• Less than 1ms (around ~0.2-0.5 depends on complexity of meshed area)
• Similiar to : https://github.com/bigos91/fastNaiveSurfaceNets
• Triangulation table is a bit different than original - different corner indexing is used to make simd stuff possible https://www.youtube.com/watch?v=fIzZdO7FxqQ&ab_channel=Bigos91

Features:

• 2 different triangulation tables (read Mesher.Arrays.cs)
• Naive version, SIMD, and SIMD multithreaded.
• No normals generated, intead they are calculated in fragment shader (ddx,ddy)
• No comment on simd stuff - if you need explanation read https://github.com/bigos91/fastNaiveSurfaceNets it is same.
• Cornermask calculations are done using SIMD stuff, 32 cubes at time (32x2x2 voxels), reusing values calculated from previous loop steps.

Problems:

• Multithreaded version sometimes gives better performance results, sometimes not. No idea why.

Limitations:

• Meshed area must have 32 voxels in Z dimension to make SIMD implementation work
• 32x32x32 volumes, but it is possible to make it working with 32xNxM (Chunk.cs), or any size if you do not use SIMD stuff.
• It is only meshing algorithm. No any king of world management, etc.

Requirements:

• Unity (2020.3 works fine, dont know about previous versions)
• CPU with SSE4.1 support (around year 2007)

Usage:

• Clone, run, open scene [FastMarchingCubes/Scenes/SampleScene],
• Disable everything what makes burst safe to make it faster :)

Resources:

• http://paulbourke.net/geometry/polygonise/ - table
• http://paulbourke.net/geometry/polygonise/table2.txt - alternative table
• https://github.com/SebLague/Marching-Cubes - corner tables
• https://github.com/Chaser324/unity-wireframe - for wireframe.

Todo:

• 16^3 size version
• maybe 64^3 size version but on AVX