michaelbratsch

The issue has been fixed with: https://github.com/scikit-image/scikit-image/commit/5ccee45db9f9556297f083bb2f174fecceb4d756

> Oh, thank you @michaelbratsch! That seems correct. I'm curious, how did you come across this and did you find the commit by simply searching? :) I was looking at...

> I tried using scipy's mentioned function but the sequence of rotation axes wasn't clear to me. I was mainly confused by the inability to define the standard z-x'-z rotation...

> Sure some combination of arguments will give same results, but it's not conceptually clear. How/Why does it make sense that a rotation of (alpha, beta, gamma) from first principles...

It is a matter of notational convention and there is not really a right and wrong. Definitely, the current implementation will work. I just have to point out, that the...

Looking at the `EuclideanTransform` from scikit-image, I realised that there is already an implementation to generate rotation matrices from Euler angles (always assumed it to be only 2D :astonished:), https://github.com/scikit-image/scikit-image/blob/fe57b47ba46e9e11c626b97a8a24cd4d2c41fba1/skimage/transform/_geometric.py#L1286-L1291...

@lagru I agree that it is a good idea to use SciPy's `Rotation.from_euler` for https://github.com/scikit-image/scikit-image/blob/fe57b47ba46e9e11c626b97a8a24cd4d2c41fba1/skimage/transform/_geometric.py#L1228 and opened a pull-request for it https://github.com/scikit-image/scikit-image/pull/7503.

> Yes! We should also include the chosen conventions in our user docs (something similar to the docs on [datatypes](https://scikit-image.org/docs/stable/user_guide/data_types.html)). Good idea. Could be a separate pull-request to update the...

> the reconstructed image become blured than the original image, why ? As stated in the documentation of the radon transform: https://scikit-image.org/docs/stable/auto_examples/transform/plot_radon_transform.html > A practical, exact implementation of the inverse...

> sorry i test a few images but for this special case its blurred. Which images did you try out that did not give you any blur? > but for...