error extract-mesh

[LHXhh]

Thanks for your wonderful work, I trained my own dataset with bakedsdf, after training process done, I can export the point cloud, though the quality is low, but the process extract-mesh always be killed, the gpu memery is enough. Could you please help me.

cmd:ns-extract-mesh --load-config outputs/bakedsdf_DJIWC_202304272258/bakedsdf/2023-04-27_225853/config.yml --output-path meshes/bakedsdf_DJIWC_202304291211.ply --bounding-box-min -2.0 -2.0 -2.0 --bounding-box-max 2.0 2.0 2.0 --resolution 4096 --marching_cube_threshold 0.001 --create_visibility_mask True

(nerfstudio) root@autodl-container-4858118aae-7ddc7131:~/autodl-tmp/sdfstudio# ns-extract-mesh --load-config outputs/bakedsdf_DJIWC_202304272258/bakedsdf/2023-04-27_225853/config.yml --output-path meshes/bakedsdf_DJIWC_202304291211.ply --bounding-box-min -2.0 -2.0 -2.0 --bounding-box-max 2.0 2.0 2.0 --resolution 4096 --marching_cube_threshold 0.001 --create_visibility_mask True scale_factor self.config.scale_factor = 0.014343904331326485 scale_factor self.config.scale_factor = 0.014343904331326485 Setting up training dataset... Caching all 144 images. Setting up evaluation dataset... Caching all 144 images. Loading latest checkpoint from load_dir ✅ Done loading checkpoint from outputs/bakedsdf_DJIWC_202304272258/bakedsdf/2023-04-27_225853/sdfstudio_models/step-000250000.ckpt Extract mesh with marching cubes and may take a while /root/miniconda3/envs/nerfstudio/lib/python3.8/site-packages/torch/nn/functional.py:4215: UserWarning: Default grid_sample and affine_grid behavior has changed to align_corners=False since 1.3.0. Please specify align_corners=True if the old behavior is desired. See the documentation of grid_sample for details. warnings.warn( 0 0 0 Killed

point cloud

[niujinshuchong]

Hi, maybe the resolution is too high, so you run out of CPU RAM. Could you try to run the command with a smaller resolution, e.g. --resolution 512

[LHXhh]

Thank you very much, I will have a try.