Richard Beare

Yet more thinking and playing around with tests of fractional voxels, and the existing test images don't trigger the problem we saw with Roman's data. However things are clearer to...

Ah, I hadn't noticed that threshold. It certainly explains why the starting derivative can be zero. I was assuming that a large starting gradient caused a jump into an illegal...

Slightly more progress - I've managed to reproduce the situation where the initial gradient is zero. It happens when the values in the speed image are very low, leading to...

OK, Less simple than I thought. The PhysicalCentralDifferenceImageFunction uses the interpolator to determine image values one voxel away along each axis. My original thought was to check whether those neighbours...

I've implemented a safe interpolator and some tests. The code is in the SafeInterpolator branch of my repo. As pointed out in #63, interpolators are used in two places, and...

The reason the fast marching leaves pixels unevaluated (besides the zero speed ones), is that it stops after reaching all the targets. There is a parameter that causes it to...

@romangrothausmann I've added yet another branch with a version allowing the interpolator to be changed by the user. The default is to keep the linear interpolator, but now the user...

On the to-do list

Hi Chris, I've been helping Bonnie and co out with this. I have what I think is a temporary fix, but hardly ideal. Assuming we want to stick with the...

Happy to have more of a play - how did you decide on the original content of Resources/python37/ ?