facebookresearch / pytorch3d

PyTorch3D is FAIR's library of reusable components for deep learning with 3D data
https://pytorch3d.org/
Other
8.71k stars 1.3k forks source link

Screened Poisson surface reconstruction #331

Open wzic opened 4 years ago

wzic commented 4 years ago

🚀 Feature

It would be nice if the screened Poisson surface reconstruction algorithm could be implemented

Motivation

Screen Poisson is a surface reconstruction algorithm that has been adopted by a lot of papers to reconstruct meshes from point clouds. Having this feature in pytorch3d would help researchers who focus on reconstructing meshes.

Thanks!

reference: K AZHDAN , M., AND HOPPE , H. Screened poisson surface reconstruction. ACM Trans. Graph. 32 (2013), 29:1–29:13. pages 8, 9

Pitch

NOTE: we only consider adding new features if they are useful for many users.

gkioxari commented 4 years ago

Hi @wzic Thanks for bringing this up! It does indeed sound interesting! This op is currently not in our immediate roadmap. However, if more users request it we will include it. Note that we do focus on differentiability in lots of our ops, so if you have a function in mind that also admits gradients that's even better! Also, if you end up implementing this and wish to include it to PyTorch3D we are more than happy to review your PR!

ucalyptus commented 4 years ago

@gkioxari this is surely needed . A differentiable mesh generation method would really help.

derrick-xwp commented 3 years ago

Agree with above comments, It is urgently needed.

kajal-puri commented 2 years ago

This feature would indeed be very helpful.

sergeyprokudin commented 2 years ago

+1!

ThumbmasWalker commented 1 year ago

+1

mradzikowski commented 1 year ago

+1 differentiable mesh generation would be really great

WenHaoDing commented 7 months ago

+1

uzgit commented 4 months ago

+1

supertan0204 commented 2 months ago

+1