Open xleEmesent opened 2 years ago
This is a bit of a bug — at the moment the MeshMeta object requires a cloud volume path even if the files are to be loaded locally, because it wants to know how to handle a root id. We should improve the logic of this requirement, as well as have a direct method to do mesh loading.
There are two workarounds right now:
1) Use any segmentation cloudpath, even if you're not going to use it to download files. Here's a publicly available one precomputed://gs://neuroglancer-public-data/kasthuri2011/ground_truth
2) Build the mesh through a method other than the meshmeta — if you can import vertices and faces, then trimesh_io.Mesh(vertices=vertices, faces=faces) will work fine. Trimesh should make this import easy, but then you don't have the extra features needed for the EM neuronal meshes. However, it's also possible that Trimesh does all you need it to do if you're coming in with an obj.
@ceesem , Thanks for your reply.
Regarding to the 1st workaround where a cloud path is required. If my mesh file has path as : /home/xue/eclipse-workspace/myproject/data/Mesh.obj
. What should the cloudpath be here? Sorry I never use the cloudpath before.
I am trying to following the example but I could not figure out right cloudpath in my case here:
mm = trimesh_io.MeshMeta(
# cv_path = "precomputed://gs://neuroglancer-public-data/kasthuri2011/ground_truth",
disk_cache_path = "test_meshes",
map_gs_to_https=True)
Do I also need to set up google cloud credentials as well? Many Thanks!!!
Hi, MeshParty team:
I am trying to load a mesh file (can open in MeshLab)
Then the following error message generated:
Anything I have missed here? Thanks in advance.