TL;DR: matplotlib 3.9 removed get_cmap from matplotlib.cm. Use `matplotlib.pyplot.get_cmap(name)` or `matplotlib.colormaps[name]` instead, or upgrade the library that does the old import.

## The error
```text
ImportError: cannot import name 'get_cmap' from 'matplotlib.cm'
```

## Fix it
1. Replace the import in your code:
   ```python
   from matplotlib import colormaps
   cmap = colormaps["viridis"]
   ```
   or keep the function style:
   ```python
   from matplotlib.pyplot import get_cmap
   ```
   Expected: no import error.
2. If the bad import lives inside a third-party package (like an old seaborn-image), upgrade that package:
   ```
   pip install -U seaborn-image
   ```
   Expected: new release uses the new path.
3. Verify:
   ```
   python -c "from matplotlib import colormaps; print(colormaps['viridis'])"
   ```
   Expected: prints the colormap object.

## When this applies
Use this when code does `from matplotlib.cm import get_cmap` on matplotlib 3.9+.

## When it does not apply
If the error is about a different name in matplotlib.cm, check the 3.9 removal list. If matplotlib itself will not import, reinstall it.

## Compatibility
matplotlib 3.9+. matplotlib.colormaps exists since 3.5, so the new code works on 3.5 through current.

## Root cause
matplotlib deprecated get_cmap/get_cmap_cls in matplotlib.cm and removed them in 3.9, consolidating colormap lookup in the colormaps registry. Code that imported from the old location breaks at import.

## Edge cases
Old tutorials are full of `plt.get_cmap`; that path still works because pyplot re-exports it. Only the direct `from matplotlib.cm import get_cmap` form breaks.