Closed linlinge closed 4 months ago
Hi Linlin, the pre-trained models do not support this, they only support estimating normals/curvature directly from point clouds (so without unoriented normals as input). You could alter the code a bit to accept 6-dimensional point inputs (xyz, unoriented normal) and train this updated model to predict the oriented normal. Although there are also methods more specialized in your problem like this one: https://cims.nyu.edu/gcl/papers/2021-Dipole.pdf
Hope that helps!
Hello! I want to use the pre-trained models to obtain a consistent normal orientation for a point cloud that already has the estimated normals. Which pre-trained model should I use?