CudaSDF

This project implements a Marching Cubes algorithm using CUDA and OpenGL for visualization.

Prerequisites

• CUDA Toolkit (11.0 or newer recommended, requires cub which is included in modern Toolkits)
• CMake (3.18+)
• C++ Compiler (MSVC, GCC, Clang)
• OpenGL libraries
• GLFW3 library
• GLAD (Used for OpenGL function loading)

Build Instructions

1. Create a build directory:

   mkdir build
   cd build

2. Configure with CMake:

   cmake ..

   Note: If CMake cannot find GLAD, you may need to download glad.c and glad.h from https://glad.dav1d.de/ and place them in the src directory or adjust CMakeLists.txt.

3. Build:

   cmake --build . --config Release

4. Run:

   ./Release/CudaSDF.exe

Project Structure

• src/main.cpp: Main entry point, OpenGL setup, and rendering loop.
• src/MarchingCubesKernels.cu: CUDA kernels for SDF generation, vertex counting/creation, and index generation.
• src/Wrappers.cu: Wrapper functions to call CUDA kernels from C++.
• src/Commons.cuh: Common data structures (SDFGrid) and helper functions.
• src/MarchingCubesTables.cuh: Lookup tables for the Marching Cubes algorithm.

Implementation Details

The implementation follows a standard GPU Marching Cubes pipeline:

1. Generate SDF: Evaluates a Signed Distance Function (sphere) on a grid.
2. Count Vertices: Checks each cell to see if it intersects the surface and counts required vertices.
3. Scan Vertices: Uses cub::DeviceScan to compute prefix sums of vertex counts, determining the memory offset for each cell's vertices.
4. Create Vertices: Interpolates vertex positions along edges and stores them in a Vertex Buffer Object (VBO) mapped from OpenGL.
5. Count Indices: Determines the number of triangle indices required for each cell.
6. Scan Indices: Computes prefix sums for index offsets.
7. Create Indices: Generates triangle topology indices into an Element Array Buffer (EBO) mapped from OpenGL.
8. Render: Draws the mesh using OpenGL.

Controls

• W: Toggle wireframe rendering
• Space: Toggle mesh animation
• C: Toggle view rotation
• M: Switch mesh extraction to Marching Cubes
• D: Switch mesh extraction to Dual Contouring
• [ / ]: Decrease / increase Dual Contouring normal smoothing angle
• B: Toggle Dual Contouring normal mode (combined vs dominant primitive gradient)
• J: Cycle Dual Contouring vertex projection iterations (0–9)
• K: Toggle Dual Contouring refinement normal mode
• L: Toggle Dual Contouring adaptive quad diagonal selection
• V: Toggle UV generation
• U: Trigger UV unwrap
• A: Trigger UV atlas packing
• P: Trigger texture projection
• T: Toggle texture array mode
• N: Toggle vertex normals mode (accurate vs SDF gradient)
• O: Export mesh to OBJ
• Esc: Exit

demo.mp4