dissectible_anatomy

NVIDIA only: Code relies on NVIDIA extensions Use left mouse button to grab and right mouse button to cut.

DissectDemo

An OpenGL 4.6 demo that lets you interactively cut/dissect a 3D mesh in real time, built on a compute-shader voxelizer, an XPBD-based soft-body solver, and ImGui-driven controls. Originally a Visual Studio project; this directory is a CMake-based distribution of the same source so it can be built on Windows (MSVC) and Linux without hand-installing any libraries into solution-relative folders.

Some rendering paths use NVIDIA’s mesh-shader (GL_NV_mesh_shader) and subgroup extensions, so a recent NVIDIA GPU with up-to-date drivers gives the best (and most complete) result.

Dependencies

Library	Windows (MSVC)	Linux
GLFW	via vcpkg	`libglfw3-dev`
GLEW	via vcpkg	`libglew-dev`
Assimp	via vcpkg	`libassimp-dev`
FreeImage	via vcpkg	`libfreeimage-dev`
GLM	via vcpkg	`libglm-dev`
Dear ImGui	fetched by CMake (FetchContent) at configure time	same

Building on Windows (MSVC)

Install vcpkg and set the VCPKG_ROOT environment variable to its location.
Install CMake 3.21+ and Visual Studio 2022 (with the “Desktop development with C++” workload).
From the repo root:
```
cmake --preset windows-msvc
cmake --build --preset windows-msvc
```
The first configure will take a few minutes while vcpkg builds GLFW, GLEW, Assimp, FreeImage, and GLM for you.
Run it:
```
cd bin
.\DissectDemo.exe
```
The build always places the executable at bin/ in the repo root (regardless of platform, generator, or Debug/Release config), and copies shaders/, assets/, and default DissectDemo.ini/imgui.ini config there too (only if not already present, so a rebuild won’t overwrite settings you’ve since changed in-app), with vcpkg copying the dependency DLLs alongside them, so the folder is self-contained.

Building on Linux

Install the dev packages, then configure and build:

sudo apt-get install libglfw3-dev libglew-dev libassimp-dev libfreeimage-dev libglm-dev libgl1-mesa-dev mesa-common-dev ninja-build
cmake --preset linux
cmake --build --preset linux
cd bin
./DissectDemo

If you’d rather not install system packages, use the linux-vcpkg preset instead (same vcpkg setup as the Windows instructions above).

Notes

The app expects shaders/ and assets/ as subdirectories of the current working directory, so run it from bin/ (see above) rather than invoking it via a path from elsewhere.
The build always places the executable at <repo>/bin/, on every platform/generator/config, specifically so that paths relative to the working directory stay predictable. This matters for the optional “virtual dissection data” scenes in SdfGridBuilder.h (the vh high and vh female datasets), which look for an external dataset at ../virtual dissection data/. Since the executable always runs from <repo>/bin/, that path always resolves to <repo>/virtual dissection data/ – a sibling of bin/, src/, shaders/, etc., regardless of platform.
CMakePresets.json defines windows-msvc, linux, and linux-vcpkg configure/build presets. Run cmake --list-presets to see them.

Optional: the “vh high” / “vh female” dissection data

These two scenes aren’t required to run the default armadillo demo, and the underlying imagery is too large to track in git (GitHub caps normal commits at 100MB/file), so it’s distributed as assets on this repo’s dissection-data-v1 release instead. Fetch and place it automatically with:

.\scripts\fetch-dissection-data.ps1

This requires the GitHub CLI (gh), authenticated via gh auth login – necessary while this repo is private, since a plain download link won’t work without authentication. The script places the data at <repo>/virtual dissection data/vh high/ and vh female/, matching what SdfGridBuilder.h expects, and skips any files already present so re-running it is safe.