Hi thank you for making the code available.
I was able to set it up and partially run it on the dataset you have made available. Following the steps as shown:
- Ran
InstantNGP to obtain the coarse shape for the DURIAN data sample, save the mesh and the point cloud (PCL)
- Update
data_args.py and run main.py to make the GUI pop up.
- Start the training and and stop it once the result looks converged and the loss is no longer fluctuating too much.
- Save the checkpoint and run sample_patches.
- Run
patch_matching_and_quilting.py by updated the path to the workspace to obtain texture.npz
I do need to mention that I have to close the GUI after step 4, as patch_matching_and_quilting.py requires almost my entire RAM to run.
Then I reopen the GUI to do the next steps,
I was able to complete these steps however when I run load_synthesis, in the terminal it finishes the required code to run in the background but the output display of the GUI is blank.
Same problem happens for load_shape as well. Kindly let me know if I have missed something or is there a way to make the patch_matching_and_quilting.py less expensive in terms of storage.
System Configuration:
Ubuntu 24.04 16GB RAM
Intel i7 10th Gen, 12 cores
NVIDIA 1660 Laptop GPU
Python 3.9 + PyTorch 2.5.1
CUDA Version 12.1
There were little to no issues in setting up the code in spite of version difference. The only change I had to make in the code was the trimesh.as_open3d has been deprecated so I had change that to explicitly create an open3d and store the values of vertices, faces and normals from a Trimesh object.
Hi thank you for making the code available.
I was able to set it up and partially run it on the dataset you have made available. Following the steps as shown:
InstantNGPto obtain the coarse shape for the DURIAN data sample, save the mesh and the point cloud (PCL)data_args.pyand runmain.pyto make the GUI pop up.patch_matching_and_quilting.pyby updated the path to the workspace to obtain texture.npzI do need to mention that I have to close the GUI after step 4, as
patch_matching_and_quilting.pyrequires almost my entire RAM to run.Then I reopen the GUI to do the next steps,
I was able to complete these steps however when I run
load_synthesis, in the terminal it finishes the required code to run in the background but the output display of the GUI is blank.Same problem happens for
load_shapeas well. Kindly let me know if I have missed something or is there a way to make thepatch_matching_and_quilting.pyless expensive in terms of storage.System Configuration:
There were little to no issues in setting up the code in spite of version difference. The only change I had to make in the code was the
trimesh.as_open3dhas been deprecated so I had change that to explicitly create an open3d and store the values of vertices, faces and normals from a Trimesh object.