3
$\begingroup$

I am more or less a complete novice in the field of 3D modeling. I understand the basics of a polygon mesh and a texture; however, it was always my understanding that 3D models were, effectively, "hollow" inside the mesh, in the same way that a plastic skeleton Halloween decoration is hollow. It possesses a surface, but no internal density. This understanding has always worked well for me, as I primarily work with photogrammetry-created 3D models, for which this is undoubtedly true.

In my latest project, however, I am being asked to work with 3D models created using CT scan data and Slicer. Unlike photogrammetry images, CT scans do care about internal density and structure. I know Slicer can be used to create 3D models from CT scan images. What I'm uncertain about, however, is whether Slicer creates truly "solid" models from this data -- models which have internal density rather than simply a mesh "skin", where the "inside" of the model is programmatically distinct from the "outside" -- or whether it simply create a multilayered but still fundamentally hollow mesh (or series of meshes) based on boundaries between areas of different densities.

Is this a meaningful distinction? Am I fundamentally misunderstanding how the surface/internal geometry of 3D models works? Any advice is welcome.

(I took my best guess as to where to ask this question -- if this is outside the purview of this site, please let me know! If you know another place where I could go asking for answers, I would also appreciate that information. Thank you so much!!!)

$\endgroup$
  • $\begingroup$ It might be worth adding that we also call boundary representation models sold when they are enforced to be closed. They are still hollow though. $\endgroup$ – joojaa Jun 20 at 6:41
4
$\begingroup$

Assuming what you’re referring to is this Slicer and the models are the ones produced by its Model Maker module: it looks like it’s creating hollow, surface-only models.

screenshot from the Model Maker module in Slicer

Specifically, judging by the crunchiness of the mesh shown in the top right of this screenshot, it looks like it’s using marching cubes to create a mesh from the isosurfaces at given density thresholds in the volume.

Most 3D model formats—especially the ones used for real-time rendering—are surface-only; as graphics hardware gets more powerful, though, there’s been wider use of volume rendering, often via raymarching, to render volume data directly rather than having to go through the intermediate meshing step. One common format for volume data is OpenVDB; it’s also possible to embed such data in Alembic files.

| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ Thank you so much!!!! This helps quite a bit. $\endgroup$ – confusedArchaeologist Jun 19 at 22:07
  • 1
    $\begingroup$ The main problem with volume (voxel) data is, that with growing size of your object, you need more and more voxels (power of 3), especially when you want to maintain some precision. With BREP (boundary representation) you save memory space. So I guess for representing your archeological site you will choose BREP (some even choose a colored point cloud), while skull data of a mummy should be a voxel-based CT scan which you still can convert to a BREP (via iso-surface calculation) for a VR visualization. $\endgroup$ – Trantor Jun 21 at 16:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.