Bug description

Signal1D fails to plot the navigation axes when using large navigation dimensions (e.g., 256 × 256) with a low-bit datatype (e.g., np.float16).

The navigation pane appears blank (white) instead of showing the expected image.

Expected behaviour

The following example (with np.float64 data) works correctly:

import hyperspy.api as hs
import numpy as np

# Generate sample data
data = np.random.random((256, 256, 10))

# This works fine
s1 = hs.signals.Signal1D(data=data)
s1.plot()

This successfully plots the navigation pane (showing a static noise image).

To reproduce

Changing the datatype to np.float16 causes the navigation pane to render incorrectly:

# This shows a white navigation image
s2 = hs.signals.Signal1D(data=data)
s2.change_dtype(np.float16)
s2.plot()

The resulting plot shows a white navigation pane:

Python environment:

• HyperSpy version: 2.3.0
• Python version: 3.13

Additional context

Interestingly, the issue appears when the navigation size reaches the limit of what can be represented in np.float16. For example, changing the data shape to (256, 255, 10) allows s2.plot() to render correctly even with datatype changed to np.float16.