Hi,
Thank you for sharing your great work! I have a question regarding the way the SDF ground truth (GT) is generated in your code. Specifically, I noticed the following snippet:
// class PointCloud in dataio.py (line 435)
sdf = np.zeros((total_samples, 1)) # on-surface = 0
sdf[self.on_surface_points:, :] = -1 # off-surface = -1
Typically, an SDF is defined as a signed distance function, where off-surface points have positive values for the exterior and negative values for the interior. However, in this implementation, the off-surface SDF GT values are set to -1. Could you explain the reasoning behind this choice? Is there a specific consideration or benefit to setting off-surface values to -1 rather than using the actual signed distance?
Looking forward to your insights. Thanks in advance!
Hi,
Thank you for sharing your great work! I have a question regarding the way the SDF ground truth (GT) is generated in your code. Specifically, I noticed the following snippet:
Typically, an SDF is defined as a signed distance function, where off-surface points have positive values for the exterior and negative values for the interior. However, in this implementation, the off-surface SDF GT values are set to -1. Could you explain the reasoning behind this choice? Is there a specific consideration or benefit to setting off-surface values to -1 rather than using the actual signed distance?
Looking forward to your insights. Thanks in advance!