diff --git a/napari/layers/base/base.py b/napari/layers/base/base.py
index 54d54045c8f..cd3d36dc9d7 100644
--- a/napari/layers/base/base.py
+++ b/napari/layers/base/base.py
@@ -869,9 +869,7 @@ def thumbnail(self, thumbnail):
         if 0 in thumbnail.shape:
             thumbnail = np.zeros(self._thumbnail_shape, dtype=np.uint8)
         if thumbnail.dtype != np.uint8:
-            with warnings.catch_warnings():
-                warnings.simplefilter("ignore")
-                thumbnail = convert_to_uint8(thumbnail)
+            thumbnail = convert_to_uint8(thumbnail)
 
         padding_needed = np.subtract(self._thumbnail_shape, thumbnail.shape)
         pad_amounts = [(p // 2, (p + 1) // 2) for p in padding_needed]
diff --git a/napari/layers/image/image.py b/napari/layers/image/image.py
index bc69f117304..3806628b481 100644
--- a/napari/layers/image/image.py
+++ b/napari/layers/image/image.py
@@ -901,12 +901,9 @@ def _update_thumbnail(self):
         )
         zoom_factor = tuple(new_shape / image.shape[:2])
         if self.rgb:
-            # warning filter can be removed with scipy 1.4
-            with warnings.catch_warnings():
-                warnings.simplefilter("ignore")
-                downsampled = ndi.zoom(
-                    image, zoom_factor + (1,), prefilter=False, order=0
-                )
+            downsampled = ndi.zoom(
+                image, zoom_factor + (1,), prefilter=False, order=0
+            )
             if image.shape[2] == 4:  # image is RGBA
                 colormapped = np.copy(downsampled)
                 colormapped[..., 3] = downsampled[..., 3] * self.opacity
@@ -923,12 +920,9 @@ def _update_thumbnail(self):
                     alpha = np.full(downsampled.shape[:2] + (1,), self.opacity)
                 colormapped = np.concatenate([downsampled, alpha], axis=2)
         else:
-            # warning filter can be removed with scipy 1.4
-            with warnings.catch_warnings():
-                warnings.simplefilter("ignore")
-                downsampled = ndi.zoom(
-                    image, zoom_factor, prefilter=False, order=0
-                )
+            downsampled = ndi.zoom(
+                image, zoom_factor, prefilter=False, order=0
+            )
             low, high = self.contrast_limits
             downsampled = np.clip(downsampled, low, high)
             color_range = high - low
diff --git a/napari/layers/utils/layer_utils.py b/napari/layers/utils/layer_utils.py
index 4fb62832324..2935854abe3 100644
--- a/napari/layers/utils/layer_utils.py
+++ b/napari/layers/utils/layer_utils.py
@@ -278,18 +278,20 @@ def segment_normal(a, b, p=(0, 0, 1)):
 
 def convert_to_uint8(data: np.ndarray) -> np.ndarray:
     """
-    Convert array content to uint8.
+    Convert array content to uint8, always returning a copy.
 
-    If all negative values are changed on 0.
+    Based on skimage.util.dtype._convert but limited to an output type uint8,
+    so should be equivalent to skimage.util.dtype.img_as_ubyte.
 
-    If values are integer and bellow 256 it is simple casting otherwise maximum value for this data type is picked
-    and values are scaled by 255/maximum type value.
+    If all negative, values are clipped to 0.
 
-    Binary images ar converted to [0,255] images.
+    If values are integers and below 256, this simply casts.
+    Otherwise the maximum value for the input data type is determined and
+    output values are proportionally scaled by this value.
 
-    float images are multiply by 255 and then casted to uint8.
+    Binary images are converted so that False -> 0, True -> 255.
 
-    Based on skimage.util.dtype.convert but limited to output type uint8
+    Float images are multiplied by 255 and then cast to uint8.
     """
     out_dtype = np.dtype(np.uint8)
     out_max = np.iinfo(out_dtype).max
@@ -302,6 +304,7 @@ def convert_to_uint8(data: np.ndarray) -> np.ndarray:
         image_out = np.multiply(data, out_max, dtype=data.dtype)
         np.rint(image_out, out=image_out)
         np.clip(image_out, 0, out_max, out=image_out)
+        image_out = np.nan_to_num(image_out, copy=False)
         return image_out.astype(out_dtype)
 
     if in_kind in "ui":