Create block.py

block.py

@@ -0,0 +1,96 @@
+from diffusers.modular_pipelines import (
+    PipelineBlock,
+    InputParam,
+    OutputParam,
+    ConfigSpec,
+)
+
+from diffusers.utils import load_image
+from PIL import Image
+from typing import Union, Tuple
+
+# copied from https://github.com/Wan-Video/Wan2.2/blob/388807310646ed5f318a99f8e8d9ad28c5b65373/wan/utils/utils.py#L136
+def best_output_size(w, h, dw, dh, expected_area):
+    # float output size
+    ratio = w / h
+    ow = (expected_area * ratio)**0.5
+    oh = expected_area / ow
+
+    # process width first
+    ow1 = int(ow // dw * dw)
+    oh1 = int(expected_area / ow1 // dh * dh)
+    assert ow1 % dw == 0 and oh1 % dh == 0 and ow1 * oh1 <= expected_area
+    ratio1 = ow1 / oh1
+
+    # process height first
+    oh2 = int(oh // dh * dh)
+    ow2 = int(expected_area / oh2 // dw * dw)
+    assert oh2 % dh == 0 and ow2 % dw == 0 and ow2 * oh2 <= expected_area
+    ratio2 = ow2 / oh2
+
+    # compare ratios
+    if max(ratio / ratio1, ratio1 / ratio) < max(ratio / ratio2,
+                                                 ratio2 / ratio):
+        return ow1, oh1
+    else:
+        return ow2, oh2
+
+class Wan225BI2VImageProcessor(PipelineBlock):
+
+    @property
+    def description(self):
+        return "default Image Processor for wan2.2 5b i2v, it resizes the image to the best output size and center-crop it"
+
+    @property
+    def inputs(self):
+        return [
+            InputParam(name="image", type_hint=Union[Image.Image, str], description= "the Image to process"),
+            InputParam(name="max_area", type_hint=int, description= "the maximum area of the Image to process")
+        ]
+    
+    @property
+    def intermediate_outputs(self):
+        return [
+            OutputParam(name="processed_image", type_hint=Image.Image, description= "the processed Image"),
+        ]
+    
+    @property
+    def expected_configs(self):
+        return [
+            ConfigSpec(name="patch_size", default=(1, 2, 2)),
+            ConfigSpec(name="vae_stride", default=(4, 16, 16)),
+        ]
+    
+    def __call__(self, components, state):
+
+        block_state = self.get_block_state(state)
+
+        if isinstance(block_state.image, str):
+            image = load_image(block_state.image).convert("RGB")
+        elif isinstance(block_state.image, Image.Image):
+            image = block_state.image
+        else:
+            raise ValueError(f"Invalid image type: {type(block_state.image)}; only support PIL Image or url string")
+        
+        ih, iw = image.height, image.width
+        dh, dw = components.patch_size[1] * components.vae_stride[1], components.patch_size[2] * components.vae_stride[2]
+        ow, oh = best_output_size(iw, ih, dw, dh, block_state.max_area)
+
+        scale = max(ow / iw, oh / ih)
+        resized_image = image.resize((round(iw * scale), round(ih * scale)), Image.LANCZOS)
+
+        # center-crop
+        x1 = (resized_image.width - ow) // 2
+        y1 = (resized_image.height - oh) // 2
+        cropped_image = resized_image.crop((x1, y1, x1 + ow, y1 + oh))
+        assert cropped_image.width == ow and cropped_image.height == oh
+
+        block_state.processed_image = cropped_image
+
+        print(f" initial image size: {image.size}")
+        print(f" processed image size: {cropped_image.size}")
+        
+
+        self.set_block_state(state, block_state)
+        return components, state
+