michaelbratsch

Looking at the code, it seems to me that the parameters `max_cumulative_cost` and `max_cost` of `find_costs` are not used at all. Furthermore, the parameter `max_coverage` is missing documentation. https://github.com/scikit-image/scikit-image/blob/707d937c43041737b562c3fb8d1fa88a5d7e13af/skimage/graph/_mcp.pyx#L395L396 The...

> As to their intended meaning, I think `max_cost` would be a cost above which the front would refuse to move to that voxel. It allows you to basically create...

I opened a merge request for the discussed changes https://github.com/scikit-image/scikit-image/blob/a854ec56aeea68f0b45c6b1ff835fdd950bcb458/skimage/graph/_mcp.pyx#L612 What still irritates me is that `find_costs` returns the internal state of `cumulative_costs` and `traceback` and not a copy. I...

@almarklein Thanks for the info. Another question, the parameter `max_coverage` is missing documentation https://github.com/scikit-image/scikit-image/blob/aa7042ccd227e7f49b73ac466542088ec5a9ad63/skimage/graph/_mcp.pyx#L396 I wanted to add some but it seems like it was only useful while developing: https://github.com/scikit-image/scikit-image/blob/aa7042ccd227e7f49b73ac466542088ec5a9ad63/skimage/graph/_mcp.pyx#L492...

I also realised that `plot_matches` has been renamed to `plot_matched_features` and is being removed for skimage 0.25 due to deprecation https://github.com/scikit-image/scikit-image/pull/7487. The tutorials https://github.com/scikit-image/skimage-tutorials still reference `plot_matches` though. If wanted...

> simply being able to fix the seed? I think it is reasonable to have random but deterministic colours. But who needs random colours which are not deterministic? I guess...

I realised there are quite some ways to specify colours in matplotlib, see https://matplotlib.org/stable/users/explain/colors/colors.html. So to know if the user passed a single color or a container with many colours,...

It is the same behaviour as one gets with matplotlib: ```python import matplotlib.pyplot as plt moon = data.moon() camera = data.camera() #plt.figure() plt.imshow(moon) #plt.figure() plt.imshow(camera) plt.show() ``` depending on the...

> Present EuclideanTransform just uses the scipy's CRS which is not consistent with skimage's present convention. Which skimage's present convention do you mean? Maybe I am missing it in the...

The same is true for `morphology.opening` and `morphology.binary_opening`. It is an issue with the underlying scipy implementation see https://github.com/scipy/scipy/issues/13991 There are also similar discussions in https://github.com/scikit-image/scikit-image/issues/1190, for example: https://github.com/scikit-image/scikit-image/issues/1190#issuecomment-572384985 Maybe...