lizhe00 / AvatarCap

Code of [ECCV 2022] "AvatarCap: Animatable Avatar Conditioned Monocular Human Volumetric Capture"
http://www.liuyebin.com/avatarcap/avatarcap.html
179 stars 22 forks source link

Trimesh.contains too slow #3

Closed lixiaoming0017 closed 2 years ago

lixiaoming0017 commented 2 years ago

cano_smpl_trimesh = trimesh.Trimesh(self.cano_smpl.posed_vertices, smpl_params.faces, process = False, use_embree = True) invalid_pts_ov = cano_smpl_trimesh.contains(invalid_pts) # [0, 1] 代码在这里执行非常慢?请问一下原因是什么呢?我看了一下invalid_pts的size有4千多万个点,有没有加速的方案的,否则跑个实验就很困难。

lizhe00 commented 2 years ago

Hello, maybe you did not install all the dependencies of trimesh, especially rtree. Please check this for the installation of full dependencies.