pyssim error with numpy 1.20
Numpy removed support for numpy.float
in version 1.20, which is used by pyssim
. This library is only used in tests and there are no better alternatives (which are lightweight at least).
I have submitted a pull request to the repository but we should consider either
- Limit numpy's version
- Write our own version in the code (which I think is easy)
- Rewrite test to use a different comparison function