The core idea is applying a B-Rep-style topology layer on top of a 3D mesh representation.
Instead of treating the mesh as the model itself, the mesh is just the geometric carrier. On top of that, there’s an explicit topology layer that tracks:
- Faces as structured surface regions
- Edges as shared boundaries
- Vertex connectivity and adjacency
- Ownership relationships between elements
The mesh handles rendering and geometric evaluation.
The topology layer handles structure.
This separation makes it possible to:
- Perform parametric edits against stable topological entities
- Run boolean operations while preserving structural relationships
- Maintain a consistent model even as geometry changes
It’s been an interesting exercise in bridging two worlds: the performance and simplicity of mesh engines, and the structured consistency of boundary representation modeling.
If you’re into CAD kernels, computational geometry, or browser-based engineering tools, I’d love feedback especially on architectural tradeoffs when layering topology over meshes.
Hey HN,
I’ve been building BREP.io, a parametric CAD system that runs entirely in the browser.
https://BREP.io
https://github.com/mmiscool/BREP.io
The core idea is applying a B-Rep-style topology layer on top of a 3D mesh representation.
Instead of treating the mesh as the model itself, the mesh is just the geometric carrier. On top of that, there’s an explicit topology layer that tracks: - Faces as structured surface regions - Edges as shared boundaries - Vertex connectivity and adjacency - Ownership relationships between elements
The mesh handles rendering and geometric evaluation. The topology layer handles structure.
This separation makes it possible to: - Perform parametric edits against stable topological entities - Run boolean operations while preserving structural relationships - Maintain a consistent model even as geometry changes
It’s been an interesting exercise in bridging two worlds: the performance and simplicity of mesh engines, and the structured consistency of boundary representation modeling.
If you’re into CAD kernels, computational geometry, or browser-based engineering tools, I’d love feedback especially on architectural tradeoffs when layering topology over meshes.
Happy to answer questions about how it works.
- mmiscool