Fix output_padding ignored if output_resize_to_target_size was not set

inpaint_cropandstitch.py

@@ -445,6 +445,44 @@ class CPUProcessorLogic(ProcessorLogic):
                     overflow = (new_h - image_h) // 2
                     new_y = -overflow
 
+        # Step 3b: When not resizing output, ensure dimensions are at least target dimensions
+        # This ensures output_padding works correctly even without resize (Option A: expand context, keep centered)
+        if not resize_output:
+            if new_w < target_w:
+                grow_w = target_w - new_w
+                new_x -= grow_w // 2
+                new_w = target_w
+                # Recalculate bounds
+                if new_x < 0:
+                    shift = -new_x
+                    if new_x + new_w + shift <= image_w:
+                        new_x += shift
+                    else:
+                        new_x = -((new_w - image_w) // 2)
+                elif new_x + new_w > image_w:
+                    overflow = new_x + new_w - image_w
+                    if new_x - overflow >= 0:
+                        new_x -= overflow
+                    else:
+                        new_x = -((new_w - image_w) // 2)
+            if new_h < target_h:
+                grow_h = target_h - new_h
+                new_y -= grow_h // 2
+                new_h = target_h
+                # Recalculate bounds
+                if new_y < 0:
+                    shift = -new_y
+                    if new_y + new_h + shift <= image_h:
+                        new_y += shift
+                    else:
+                        new_y = -((new_h - image_h) // 2)
+                elif new_y + new_h > image_h:
+                    overflow = new_y + new_h - image_h
+                    if new_y - overflow >= 0:
+                        new_y -= overflow
+                    else:
+                        new_y = -((new_h - image_h) // 2)
+
         # Step 4: Grow the image to accommodate the new context area
         up_padding, down_padding, left_padding, right_padding = 0, 0, 0, 0
 
@@ -969,6 +1007,44 @@ class GPUProcessorLogic(ProcessorLogic):
                     overflow = (new_h - image_h) // 2
                     new_y = -overflow
 
+        # Step 3b: When not resizing output, ensure dimensions are at least target dimensions
+        # This ensures output_padding works correctly even without resize (Option A: expand context, keep centered)
+        if not resize_output:
+            if new_w < target_w:
+                grow_w = target_w - new_w
+                new_x -= grow_w // 2
+                new_w = target_w
+                # Recalculate bounds
+                if new_x < 0:
+                    shift = -new_x
+                    if new_x + new_w + shift <= image_w:
+                        new_x += shift
+                    else:
+                        new_x = -((new_w - image_w) // 2)
+                elif new_x + new_w > image_w:
+                    overflow = new_x + new_w - image_w
+                    if new_x - overflow >= 0:
+                        new_x -= overflow
+                    else:
+                        new_x = -((new_w - image_w) // 2)
+            if new_h < target_h:
+                grow_h = target_h - new_h
+                new_y -= grow_h // 2
+                new_h = target_h
+                # Recalculate bounds
+                if new_y < 0:
+                    shift = -new_y
+                    if new_y + new_h + shift <= image_h:
+                        new_y += shift
+                    else:
+                        new_y = -((new_h - image_h) // 2)
+                elif new_y + new_h > image_h:
+                    overflow = new_y + new_h - image_h
+                    if new_y - overflow >= 0:
+                        new_y -= overflow
+                    else:
+                        new_y = -((new_h - image_h) // 2)
+
         # Step 4: Grow the image to accommodate the new context area
         up_padding, down_padding, left_padding, right_padding = 0, 0, 0, 0
 
@@ -1003,9 +1079,9 @@ class GPUProcessorLogic(ProcessorLogic):
 
         # Fill the new extended areas with the edge values of the image
         if up_padding > 0:
-            expanded_image[:, :, :up_padding, left_padding:left_padding + image_w] = image[:, :, 0:1, left_padding:left_padding + image_w].repeat(1, 1, up_padding, 1)
+            expanded_image[:, :, :up_padding, left_padding:left_padding + image_w] = expanded_image[:, :, up_padding:up_padding + 1, left_padding:left_padding + image_w].repeat(1, 1, up_padding, 1)
         if down_padding > 0:
-            expanded_image[:, :, -down_padding:, left_padding:left_padding + image_w] = image[:, :, -1:, left_padding:left_padding + image_w].repeat(1, 1, down_padding, 1)
+            expanded_image[:, :, -down_padding:, left_padding:left_padding + image_w] = expanded_image[:, :, up_padding + image_h - 1:up_padding + image_h, left_padding:left_padding + image_w].repeat(1, 1, down_padding, 1)
         if left_padding > 0:
             expanded_image[:, :, up_padding:up_padding + image_h, :left_padding] = expanded_image[:, :, up_padding:up_padding + image_h, left_padding:left_padding+1].repeat(1, 1, 1, left_padding)
         if right_padding > 0:

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "comfyui-inpaint-cropandstitch"
 description = "The '✂️ Inpaint Crop' and '✂️ Inpaint Stitch' nodes enable inpainting only on masked area very easily: crop the image around the masked area with the Crop node, then use any standard workflow for sampling, then connect the sampled image to the Stitch node, which will put it back in place in the original image. These nodes enable faster sampling of smaller areas and take care of downsampling and upsampling to fit specific model and resource needs."
-version = "3.0.1"
+version = "3.0.2"
 license = { file = "LICENSE" }
 
 [project.urls]