Automatic aspect ratio calculation

src/seaborn_image/_grid.py

@@ -43,8 +43,8 @@ class ImageGrid:
         Number of columns to display. Defaults to None.
     height : int or float, optional
         Size of the individual images. Defaults to 3.
-    aspect : int or float, optional
-        Aspect ratio of individual images. Defaults to 1.
+    aspect : int, float or 'auto', optional
+        Aspect ratio of individual images, when set to 'auto', it calculates the aspect ratio of the images passed. Defaults to 'auto'.
     cmap : str or `matplotlib.colors.Colormap` or list, optional
         Image colormap. If input data is a list of images,
         `cmap` can be a list of colormaps. Defaults to None.
@@ -290,7 +290,8 @@ class ImageGrid:
         ...             [pol, pl, retina],
         ...             map_func=[gaussian, median, hessian],
         ...             dx=[15, 100, None],
-        ...             units="nm")
+        ...             units="nm",
+        ...             aspect=1)
 
     Change colorbar orientation
 
@@ -321,7 +322,7 @@ def __init__(
         map_func_kw=None,
         col_wrap=None,
         height=3,
-        aspect=1,
+        aspect="auto",
         cmap=None,
         robust=False,
         perc=(2, 98),
@@ -365,6 +366,10 @@ def __init__(
                         len(map_func) if len(map_func) >= len(data) else len(data)
                     )
 
+            if aspect == "auto":
+                aspect_ratios = [img.shape[1] / img.shape[0] for img in data]
+                aspect = min(aspect_ratios)
+
         elif not isinstance(data, np.ndarray):
             raise ValueError("image data must be a list of images or a 3d or 4d array.")
 
@@ -385,6 +390,9 @@ def __init__(
                 # no of columns should now be len of map_func list
                 col_wrap = len(map_func) if col_wrap is None else col_wrap
 
+            if aspect == "auto":
+                aspect = data.shape[1] / data.shape[0]
+
         elif data.ndim in [3, 4]:
             if data.ndim == 4 and data.shape[-1] not in [1, 3, 4]:
                 raise ValueError(
@@ -418,6 +426,12 @@ def __init__(
 
             _nimages = len(slices)
 
+            if aspect == "auto":
+                # Select axis where width and height are
+                height_idx, width_idx = [i for i in range(data.ndim) if i != (axis % data.ndim)][:2]
+
+                aspect = data.shape[width_idx] / data.shape[height_idx]
+
             # ---- 3D or 4D image with an individual map_func ----
             map_func_type = self._check_map_func(map_func, map_func_kw)
             # raise a ValueError if a list of map_func is provided for 3d image
@@ -765,7 +779,7 @@ def rgbplot(
     *,
     col_wrap=3,
     height=3,
-    aspect=1,
+    aspect="auto",
     cmap=None,
     alpha=None,
     origin=None,
@@ -794,8 +808,8 @@ def rgbplot(
         Number of columns to display. Defaults to 3.
     height : int or float, optional
         Size of the individual images. Defaults to 3.
-    aspect : int or float, optional
-        Aspect ratio of individual images. Defaults to 1.
+    aspect : int, float or 'auto', optional
+        Aspect ratio of individual images, when set to 'auto', it calculates the aspect ratio of the images passed. Defaults to 'auto'.
     cmap : str or `matplotlib.colors.Colormap` or list, optional
         Image colormap or a list of colormaps. Defaults to None.
     alpha : float or array-like, optional
@@ -975,8 +989,8 @@ class ParamGrid(object):
         is not None and `row` is None. Defaults to None.
     height : int or float, optional
         Size of the individual images. Defaults to 3.
-    aspect : int or float, optional
-        Aspect ratio of individual images. Defaults to 1.
+    aspect : int, float or 'auto', optional
+        Aspect ratio of individual images, when set to 'auto', it calculates the aspect ratio of the images passed. Defaults to 'auto'.
     cmap : str or `matplotlib.colors.Colormap`, optional
         Image colormap. Defaults to None.
     alpha : float or array-like, optional
@@ -1113,7 +1127,7 @@ def __init__(
         col=None,
         col_wrap=None,
         height=3,
-        aspect=1,
+        aspect="auto",
         cmap=None,
         alpha=None,
         origin=None,
@@ -1173,6 +1187,9 @@ def __init__(
             ncol = col_wrap
             nrow = int(np.ceil(len(kwargs[f"{col}"]) / col_wrap))
 
+        if aspect == "auto":
+            aspect = data.shape[1]/data.shape[0]
+
         # Calculate the base figure size
         figsize = (ncol * height * aspect, nrow * height)
 

tests/test_grid.py

@@ -871,7 +871,28 @@ def test_figure_size(self):
         )
         np.testing.assert_array_equal(g4.fig.get_size_inches(), (3 * 2 * 1.5, 2 * 2))
         plt.close()
+
+    def test_auto_aspect(self):
+        imgsize0 = (10, 10)
+        g0 = isns.ImageGrid([np.zeros(imgsize0) for i in range(10)], aspect='auto')
+        assert np.isclose(imgsize0[1]/imgsize0[0], g0.aspect)
+        plt.close()
 
+        imgsize1 = (10, 5)
+        g1 = isns.ImageGrid([np.zeros(imgsize1) for i in range(10)], aspect='auto')
+        assert np.isclose(imgsize1[1]/imgsize1[0], g1.aspect)
+        plt.close()
+
+        imgsize2 = (5, 10)
+        g2 = isns.ImageGrid([np.zeros(imgsize2) for i in range(10)], aspect='auto')
+        assert np.isclose(imgsize2[1]/imgsize2[0], g2.aspect)
+        plt.close()
+
+        imglist = [np.zeros(imgsize0) for i in range(4)] + [np.zeros(imgsize1) for i in range(4)] + [np.zeros(imgsize2) for i in range(4)]
+        g3 = isns.ImageGrid(imglist, aspect='auto')
+        assert np.isclose(min([imgsize0[1]/imgsize0[0], imgsize1[1]/imgsize1[0], imgsize2[1]/imgsize2[0]]), g3.aspect)
+        plt.close()
+
     def test_vmin_vmax(self):
         g = isns.ImageGrid(cells, vmin=0.5, vmax=0.75)
         for ax in g.axes.ravel():