The ldif's discuss from google

Visualization with qview: scaling issue

Hi, we're trying to visualize ellipsoids overlayed on a mesh (as in Fig 5 in the paper). But the scale of the mesh that we put into qview is not consistent with the scale of the ellipsoids. How can we scale the mesh so they're on the same scale?

For example, when visualizing ldif/ldif2mesh/test-ldif.txt with ldif/ldif2mesh/test-ldif-output.ply we notice their scales are inconsistent. In fact the ellipsoids are much smaller than the mesh (only appearing as a dot on the left):

When visualizing ldif/ldif2mesh/test-ldif.txt with ldif/ldif2mesh/test-ldif.ply however we have them correctly displayed on the same scale:

We've tried to normalize test-ldif-output.ply to a unit sphere but still the scales are inconsistent:

I would imagine ldif/ldif2mesh/test-ldif.ply was scaled/translated from ldif/ldif2mesh/test-ldif-output.ply. Where can I find the script to do this?

test or val issue

I 've already trained the model ,but when i want to test it,i failed a lot time

CalledProcessError Traceback (most recent call last)
~/ldif/ldif/inference/predict.py in _grid_eval_cuda(self, sif_vector, resolution, extent)
845 try:
--> 846 cmd_result = sp.check_output(cmd, shell=True)
847 log.info(cmd_result.decode('utf-8').replace('\n', ''))

~/anaconda3/lib/python3.7/subprocess.py in check_output(timeout, *popenargs, **kwargs)
410 return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
--> 411 **kwargs).stdout
412

~/anaconda3/lib/python3.7/subprocess.py in run(input, capture_output, timeout, check, *popenargs, **kwargs)
511 raise CalledProcessError(retcode, process.args,
--> 512 output=stdout, stderr=stderr)
513 return CompletedProcess(process.args, retcode, stdout, stderr)

CalledProcessError: Command 'CUDA_VISIBLE_DEVICES=1 /home/gemengyuan/ldif/ldif/ldif2mesh/ldif2mesh /tmp/tmp0e9rowkf/ldif.txt /home/gemengyuan/ldif/ldif/ldif2mesh/extracted.occnet /tmp/tmp0e9rowkf/grid.grd -resolution 256' returned non-zero exit status 35.

During handling of the above exception, another exception occurred:

ValueError Traceback (most recent call last)
in
1 embedding = encoder.run_example(e)
----> 2 mesh = decoder.extract_mesh(embedding, resolution=256)
3 gaps_util.mshview(mesh)

~/ldif/ldif/inference/predict.py in extract_mesh(self, sif_vectors, resolution, extent, return_success, world2local)
959 extent,
960 extract_parts=False,
--> 961 world2local=world2local)
962 grid_out_time = time.time()
963 log.verbose(f'Grid eval time: {grid_out_time - extract_start_time}')

~/ldif/ldif/inference/predict.py in _grid_eval(self, sif_vector, resolution, extent, extract_parts, world2local)
886 log.verbose('Evaluating SDF grid for mesh.')
887 if self.use_inference_kernel and not extract_parts:
--> 888 return self._grid_eval_cuda(sif_vector, resolution, extent)
889 if extract_parts or world2local:
890 log.warning('Part extraction and world2local are not supported with the'

~/ldif/ldif/inference/predict.py in _grid_eval_cuda(self, sif_vector, resolution, extent)
867 'possible.')
868 else:
--> 869 raise ValueError(f'Unrecognized error code {e.returncode} occurred'
870 f' during inference kernel evaluation: {e.output}')
871

ValueError: Unrecognized error code 35 occurred during inference kernel evaluation: b'GPUCheckOk Failure: CUDA driver version is insufficient for CUDA runtime version ldif2mesh.cu 985\n'