I am still planning the fix for another edge cases I discovered earlier in the year that led to technically nonmanifold map models. The theory is figured out but the implementation is not in — yet. Sorry to keep you guys waiting, I have been very busy recently!
When there is a "dip" down to Z=0 at just the right width to create a single edge along the bottom, an edge is created that has more than 2 faces attached. In this situation, the edge is shared by both sides of the "valley" which is 4 faces, 2 faces on each side. (click the image to see the animated version, the nonmanifold edge is highlighted in yellow)
It's obvious to where the interior volumes of the mesh are and the face normals are still correct, but 3D mesh software such as Blender and slicers are strict and mark it as nonmanifold. Slicing the model is successful but mesh softwares warn about this nonmanifold edge and using the automated mesh repair built into them has unpredictable results.
This is a relatively minor issue and occurs only on models with thin slivers along the coast, but I want to get it right before releasing new/improved maps based on this new third iteration (V3) of the process — only the best! I call this V3 since the current release of USA maps were made with V1 in 2023 and V2.2 in 2024. ( https://ansonliu.com/maps/release-notes/)
This type of multi 3D mesh generation with 2.5D constraints isn't being researched and developed anywhere else as far as I know (at least as open source like this). You can keep up to date on my progress on the technical side of this, by reading Chris Harding and my collaboration notes at https://github.com/ChHarding/TouchTerrain_for_CAGEO/pull/111#issuecomment-3830323784.
As always, thank you for your continued support! You guys motivate me to continue experimenting and "Making the Uncharted".
I am still planning the fix for another edge cases I discovered earlier in the year that led to technically nonmanifold map models. The theory is figured out but the implementation is not in — yet. Sorry to keep you guys waiting, I have been very busy recently!
When there is a "dip" down to Z=0 at just the right width to create a single edge along the bottom, an edge is created that has more than 2 faces attached. In this situation, the edge is shared by both sides of the "valley" which is 4 faces, 2 faces on each side. (click the image to see the animated version, the nonmanifold edge is highlighted in yellow)
It's obvious to where the interior volumes of the mesh are and the face normals are still correct, but 3D mesh software such as Blender and slicers are strict and mark it as nonmanifold. Slicing the model is successful but mesh softwares warn about this nonmanifold edge and using the automated mesh repair built into them has unpredictable results.
This is a relatively minor issue and occurs only on models with thin slivers along the coast, but I want to get it right before releasing new/improved maps based on this new third iteration (V3) of the process — only the best! I call this V3 since the current release of USA maps were made with V1 in 2023 and V2.2 in 2024. ( https://ansonliu.com/maps/release-notes/)
This type of multi 3D mesh generation with 2.5D constraints isn't being researched and developed anywhere else as far as I know (at least as open source like this). You can keep up to date on my progress on the technical side of this, by reading Chris Harding and my collaboration notes at https://github.com/ChHarding/TouchTerrain_for_CAGEO/pull/111#issuecomment-3830323784.
As always, thank you for your continued support! You guys motivate me to continue experimenting and "Making the Uncharted".