IndexError while refine_and_decimate

[ashawkey]

IndexError while refine_and_decimate

[gr3atchoi]

FYI, my environment is
nvidia/cuda:11.8.0-devel-ubuntu22.04.
(My 4090 isn't working well with CUDA 11.6)
Others are fine so far.

[ashawkey]

@gr3atchoi Sorry for the bug, I have fixed it in the latest commit!

[Gary-Qinghua]

I have a similar issue in refine_and_decimate.
The command of stage 1 like this:
CUDA_VISIBLE_DEVICES=4 python main.py ./data/room/ --workspace trial_sdf_360_room -O --data_format colmap --bound 8 --enable_cam_center --enable_cam_near_far --scale 0.2 --downscale 4 --stage 1 --iters 10000 --lambda_lap 1e-3 --lambda_normal 1e-3 --sdf

The error:
Traceback (most recent call last):
File "main.py", line 243, in
trainer.train(train_loader, valid_loader, max_epoch)
File "/home/zhangl/project/nerf2mesh-main/nerf/utils.py", line 931, in train
self.train_one_epoch(train_loader)
File "/home/zhangl/project/nerf2mesh-main/nerf/utils.py", line 1206, in train_one_epoch
self.model.refine_and_decimate()
File "/home/zhangl/miniconda3/envs/nerf2mesh/lib/python3.7/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/home/zhangl/project/nerf2mesh-main/nerf/renderer.py", line 252, in refine_and_decimate
cur_v, cur_f = decimate_and_refine_mesh(cur_v, cur_f, mask, decimate_ratio=self.opt.refine_decimate_ratio, refine_size=self.opt.refine_size, refine_remesh_size=self.opt.refine_remesh_size)
File "/home/zhangl/project/nerf2mesh-main/meshutils.py", line 199, in decimate_and_refine_mesh
m = pml.Mesh(verts, faces, f_scalar_array=mask)
pymeshlab.pmeshlab.PyMeshLabException: Error while creating mesh: the number of face quality values is different from the number of faces.

[ashawkey]

@Gary-Qinghua Hi, bound > 1 is not supported for sdf mode currently.

[Gary-Qinghua]

Thanks.
This command is in your scripts/runall_360_indoor_sdf.sh