ProjectPhysX, I found an interesting optimization for the marching-cubes algorithm today: Since vertex interpolation happens on axis-aligned edges of the unit cube, it's sufficient to interpolate in 1D instead of 3D. The faster interpolation makes the conditions for which edge to interpolate unnecessary, allowing to get rid of the edge table. That brings the implementation down to 73 lines, including the triangle table. 🖖🤠
https://github.com/ProjectPhysX/FluidX3D/commit/649fd40fa6270fbd0823a53b2a55f4194fc9510b#diff-464b1d19d4b616b9609031b48429081b2c215328d9f98bc5cbeac6b2b84fdbf3R456
Add comment