Open Wamdell opened 4 years ago
Thanks a lot, but even while I was still actively developing this it was far away from "working well". In the end it was not adding to my emotional well-being. I will try to work on this thing again, but will be occupied till mid 2021. Aside from the technical hurdles as broken multi-threading, missing mesh reduction or tesselation there are some methodical issues. E.g. there is no keeping track of perceived empty space as done in TSDF methods. Compared to TSDF based methods it is relatively expensive to add/remove triangles. Coming up with heuristics that behave well in such cases is non-trivial. + Even if I would end find heuristics that compensates methodical disadvantages, would that be exploitable for another paper?
Thank you for your contribution, and we will continue to complete mesh-based 3D reconstruction.
After reading the paper, I feel that this method will work well. I don't know when your code will be updated