fastplotlib · kushalkolar · Apr 10, 2025 · Feb 6, 2025 · Feb 22, 2025 · Feb 22, 2025
@@ -5,6 +5,7 @@
 
 import pygfx
 
+from ..utils.functions import subsample_array
 from ..graphics import LineGraphic, ImageGraphic, TextGraphic
 from ..graphics.utils import pause_events
 from ..graphics._base import Graphic
@@ -193,28 +194,10 @@ def _fpl_add_plot_area_hook(self, plot_area):
         self._plot_area.controller.enabled = True
 
     def _calculate_histogram(self, data):
-        if data.ndim > 2:
-            # subsample to max of 500 x 100 x 100,
-            # np.histogram takes ~30ms with this size on a 8 core Ryzen laptop
-            # dim0 is usually time, allow max of 500 timepoints
-            ss0 = max(1, int(data.shape[0] / 500))  # max to prevent step = 0
-            # allow max of 100 for x and y if ndim > 2
-            ss1 = max(1, int(data.shape[1] / 100))
-            ss2 = max(1, int(data.shape[2] / 100))
 
-            data_ss = data[::ss0, ::ss1, ::ss2]
-
-            hist, edges = np.histogram(data_ss, bins=self._nbins)
-
-        else:
-            # allow max of 1000 x 1000
-            # this takes ~4ms on a 8 core Ryzen laptop
-            ss0 = max(1, int(data.shape[0] / 1_000))
-            ss1 = max(1, int(data.shape[1] / 1_000))
-
-            data_ss = data[::ss0, ::ss1]
-
-            hist, edges = np.histogram(data_ss, bins=self._nbins)
+        # get a subsampled view of this array
+        data_ss = subsample_array(data, max_items=1e6)  # 1e6 is default
+        hist, edges = np.histogram(data_ss, bins=self._nbins)
 
         # used if data ptp <= 10 because event things get weird
         # with tiny world objects due to floating point error

@@ -405,3 +405,42 @@ def parse_cmap_values(
         colors = np.vstack([colormap[val] for val in norm_cmap_values])
 
         return colors
+
+
+def subsample_array(arr, max_items=1e6):
+    """
+    Subsamples an input array while preserving its relative dimensional proportions.
+
+    Parameters
+    ----------
+    arr : np.ndarray
+        Input array of any dimensionality to be subsampled.
+
+    max_items : int, optional
+        Largest bytesize allowed for the total size of the subsampled array. Default is 1e6.
+
+    Returns
+    -------
+    np.ndarray
+        subsampled version of the input array
+    """
+    shape = np.array(arr.shape)
+    total_elements = np.prod(shape)
+
+    if total_elements <= max_items:
+        return arr  # no need to subsample if already below the threshold
+
+    # relative proportions based on the shape
+    proportions = shape / shape.sum()
+
+    target_elements = min(total_elements, max_items)
+    target_shape = np.maximum((proportions * target_elements).astype(int), 1)
+
+    # keep total elements within limit
+    scale_factor = (target_elements / np.prod(target_shape)) ** (1 / len(shape))
+    target_shape = np.maximum((target_shape * scale_factor).astype(int), 1)
+
+    steps = np.ceil(shape / target_shape).astype(int)
+    slices = tuple(slice(None, None, step) for step in steps)
+
+    return arr[slices]