Apply to nodes

backend/src/packages/chaiNNer_standard/image/batch_processing/video_frame_iterator.py

@@ -10,7 +10,7 @@
 import numpy as np
 from sanic.log import logger
 
-from nodes.impl.image_utils import normalize, to_uint8
+from nodes.impl.image_utils import to_uint8
 from nodes.properties.inputs import (
     BoolInput,
     DirectoryInput,
@@ -73,7 +73,7 @@ class Writer:
 def VideoFrameIteratorFrameLoaderNode(
     img: np.ndarray, idx: int, video_dir: str, video_name: str
 ) -> Tuple[np.ndarray, int, str, str]:
-    return normalize(img), idx, video_dir, video_name
+    return img, idx, video_dir, video_name
 
 
 @batch_processing_group.register(

backend/src/packages/chaiNNer_standard/image/create_images/create_color_gray.py

@@ -31,7 +31,8 @@
                 width="Input0",
                 height="Input1",
                 channels="1",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image/create_images/create_color_rgb.py

@@ -45,7 +45,8 @@
                 width="Input0",
                 height="Input1",
                 channels="3",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image/create_images/create_color_rgba.py

@@ -52,7 +52,8 @@
                 width="Input0",
                 height="Input1",
                 channels="4",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image/create_images/create_noise.py

@@ -200,4 +200,4 @@ def create_noise_node(
                 kwargs["seed"] = (kwargs["seed"] + 1) % (2**32)
         img /= total_brightness
 
-    return np.clip(img, 0, 1)
+    return img

backend/src/packages/chaiNNer_standard/image/io/load_image.py

@@ -11,7 +11,6 @@
 
 from nodes.impl.dds.texconv import dds_to_png_texconv
 from nodes.impl.image_formats import get_opencv_formats, get_pil_formats
-from nodes.impl.image_utils import normalize
 from nodes.properties.inputs import ImageFileInput
 from nodes.properties.outputs import DirectoryOutput, FileNameOutput, LargeImageOutput
 from nodes.utils.utils import get_h_w_c, split_file_path
@@ -146,6 +145,4 @@ def load_image_node(path: str) -> Tuple[np.ndarray, str, str]:
             f'The image "{path}" you are trying to read cannot be read by chaiNNer.'
         )
 
-    img = normalize(img)
-
     return img, dirname, basename

backend/src/packages/chaiNNer_standard/image_adjustment/adjustments/brightness_and_contrast.py

@@ -64,4 +64,4 @@ def brightness_and_contrast_node(
             axis=2,
         )
 
-    return np.clip(img, 0, 1)
+    return img

backend/src/packages/chaiNNer_standard/image_adjustment/adjustments/color_levels.py

@@ -111,4 +111,4 @@ def color_levels_node(
         out_white_all - out_black_all  # type: ignore
     ) + out_black_all
 
-    return np.clip(img, 0, 1)
+    return img

backend/src/packages/chaiNNer_standard/image_adjustment/adjustments/invert.py

@@ -15,7 +15,7 @@
     description="Inverts all colors in an image.",
     icon="MdInvertColors",
     inputs=[ImageInput()],
-    outputs=[ImageOutput(image_type="Input0")],
+    outputs=[ImageOutput(image_type="Input0", assume_normalized=True)],
 )
 def invert_node(img: np.ndarray) -> np.ndarray:
     c = get_h_w_c(img)[2]

backend/src/packages/chaiNNer_standard/image_adjustment/adjustments/opacity.py

@@ -27,7 +27,13 @@
             unit="%",
         ),
     ],
-    outputs=[ImageOutput(image_type=expression.Image(size_as="Input0"), channels=4)],
+    outputs=[
+        ImageOutput(
+            image_type=expression.Image(size_as="Input0"),
+            channels=4,
+            assume_normalized=True,
+        )
+    ],
 )
 def opacity_node(img: np.ndarray, opacity: float) -> np.ndarray:
     """Apply opacity adjustment to alpha channel"""

backend/src/packages/chaiNNer_standard/image_adjustment/adjustments/threshold_adaptive.py

@@ -3,7 +3,7 @@
 import cv2
 import numpy as np
 
-from nodes.impl.image_utils import normalize, to_uint8
+from nodes.impl.image_utils import to_uint8
 from nodes.properties.inputs import (
     AdaptiveMethodInput,
     AdaptiveThresholdInput,
@@ -61,4 +61,4 @@ def adaptive_threshold_node(
         c,
     )
 
-    return normalize(result)
+    return result

backend/src/packages/chaiNNer_standard/image_channel/all/combine_rgba.py

@@ -32,6 +32,7 @@
                 height="Input0.height & Input1.height & Input2.height & match Input3 { Image as i => i.height, _ => any }",
             ),
             channels=4,
+            assume_normalized=True,
         ).with_never_reason(
             "The input channels have different sizes but must all be the same size."
         )

backend/src/packages/chaiNNer_standard/image_channel/all/rgba_separate.py

@@ -23,16 +23,28 @@
     inputs=[ImageInput()],
     outputs=[
         ImageOutput(
-            "R Channel", image_type=expression.Image(size_as="Input0"), channels=1
+            "R Channel",
+            image_type=expression.Image(size_as="Input0"),
+            channels=1,
+            assume_normalized=True,
         ).with_id(2),
         ImageOutput(
-            "G Channel", image_type=expression.Image(size_as="Input0"), channels=1
+            "G Channel",
+            image_type=expression.Image(size_as="Input0"),
+            channels=1,
+            assume_normalized=True,
         ).with_id(1),
         ImageOutput(
-            "B Channel", image_type=expression.Image(size_as="Input0"), channels=1
+            "B Channel",
+            image_type=expression.Image(size_as="Input0"),
+            channels=1,
+            assume_normalized=True,
         ).with_id(0),
         ImageOutput(
-            "A Channel", image_type=expression.Image(size_as="Input0"), channels=1
+            "A Channel",
+            image_type=expression.Image(size_as="Input0"),
+            channels=1,
+            assume_normalized=True,
         ),
     ],
 )

backend/src/packages/chaiNNer_standard/image_channel/transparency/transparency_merge.py

@@ -27,6 +27,7 @@
                 height="Input0.height & Input1.height",
             ),
             channels=4,
+            assume_normalized=True,
         ).with_never_reason(
             "The RGB and alpha channels have different sizes but must have the same size."
         )

backend/src/packages/chaiNNer_standard/image_channel/transparency/transparency_split.py

@@ -23,11 +23,13 @@
             "RGB Channels",
             image_type=expression.Image(size_as="Input0"),
             channels=3,
+            assume_normalized=True,
         ),
         ImageOutput(
             "Alpha Channel",
             image_type=expression.Image(size_as="Input0"),
             channels=1,
+            assume_normalized=True,
         ),
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/border/create_border.py

@@ -27,7 +27,8 @@
                 width="Input0.width + Input2 * 2",
                 height="Input0.height + Input2 * 2",
                 channels="BorderType::getOutputChannels(Input1, Input0.channels)",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/border/create_edges.py

@@ -30,7 +30,8 @@
                 width="Input0.width + Input3 + Input4",
                 height="Input0.height + Input2 + Input5",
                 channels="BorderType::getOutputChannels(Input1, Input0.channels)",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/crop/crop_border.py

@@ -27,7 +27,8 @@
                 width="(Input0.width - Input1 * 2) & int(1..)",
                 height="(Input0.height - Input1 * 2) & int(1..)",
                 channels_as="Input0",
-            )
+            ),
+            assume_normalized=True,
         ).with_never_reason(
             "The cropped area would result in an image with no width or no height."
         )

backend/src/packages/chaiNNer_standard/image_dimension/crop/crop_content.py

@@ -38,7 +38,8 @@
                         channels: Input0.channels
                     }
                 }
-                """
+                """,
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/crop/crop_edges.py

@@ -28,7 +28,8 @@
                 width="(Input0.width - (Input2 + Input3)) & int(1..)",
                 height="(Input0.height - (Input1 + Input4)) & int(1..)",
                 channels_as="Input0",
-            )
+            ),
+            assume_normalized=True,
         ).with_never_reason(
             "The cropped area would result in an image with no width or no height."
         )

backend/src/packages/chaiNNer_standard/image_dimension/crop/crop_offsets.py

@@ -28,7 +28,8 @@
                 width="min(Input4, Input0.width - Input2) & int(1..)",
                 height="min(Input3, Input0.height - Input1) & int(1..)",
                 channels_as="Input0",
-            )
+            ),
+            assume_normalized=True,
         ).with_never_reason(
             "The cropped area would result in an image with no width or no height."
         )

backend/src/packages/chaiNNer_standard/image_dimension/resize/resize_factor.py

@@ -37,7 +37,8 @@
                 width="max(1, int & round(Input0.width * Input1 / 100))",
                 height="max(1, int & round(Input0.height * Input1 / 100))",
                 channels_as="Input0",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/resize/resize_resolution.py

@@ -31,7 +31,8 @@
                 width="Input1",
                 height="Input2",
                 channels="Input0.channels",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/resize/resize_to_side.py

@@ -163,7 +163,8 @@ def compareCondition(b: uint): bool {
                     height: outSize.height,
                     channels: Input0.channels
                 }
-                """
+                """,
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_dimension/resize/tile_fill.py

@@ -27,7 +27,8 @@
                 width="Input1",
                 height="Input2",
                 channels="Input0.channels",
-            )
+            ),
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_filter/blur/box_blur.py

@@ -47,6 +47,4 @@ def box_blur_node(
         kernel[:, -1] *= x_d
 
     # Linear filter with reflected padding
-    return np.clip(
-        cv2.filter2D(img, -1, kernel, borderType=cv2.BORDER_REFLECT_101), 0, 1
-    )
+    return cv2.filter2D(img, -1, kernel, borderType=cv2.BORDER_REFLECT_101)

backend/src/packages/chaiNNer_standard/image_filter/blur/gauss_blur.py

@@ -29,6 +29,4 @@ def gaussian_blur_node(
     if sigma_x == 0 and sigma_y == 0:
         return img
     else:
-        return np.clip(
-            cv2.GaussianBlur(img, (0, 0), sigmaX=sigma_x, sigmaY=sigma_y), 0, 1
-        )
+        return cv2.GaussianBlur(img, (0, 0), sigmaX=sigma_x, sigmaY=sigma_y)

backend/src/packages/chaiNNer_standard/image_filter/blur/median_blur.py

@@ -3,7 +3,7 @@
 import cv2
 import numpy as np
 
-from nodes.impl.image_utils import normalize, to_uint8
+from nodes.impl.image_utils import to_uint8
 from nodes.properties.inputs import ImageInput, NumberInput
 from nodes.properties.outputs import ImageOutput
 
@@ -29,8 +29,6 @@ def median_blur_node(
         return img
     else:
         if radius < 3:
-            blurred = cv2.medianBlur(img, 2 * radius + 1)
+            return cv2.medianBlur(img, 2 * radius + 1)
         else:  # cv2 requires uint8 for kernel size (2r+1) > 5
-            blurred = cv2.medianBlur(to_uint8(img, normalized=True), 2 * radius + 1)
-
-        return normalize(blurred)
+            return cv2.medianBlur(to_uint8(img, normalized=True), 2 * radius + 1)

backend/src/packages/chaiNNer_standard/image_filter/correction/avg_color_fix.py

@@ -93,4 +93,4 @@ def average_color_fix_node(
     if alpha is not None:
         result = np.concatenate([result, alpha], axis=2)
 
-    return np.clip(result, 0, 1)
+    return result

backend/src/packages/chaiNNer_standard/image_filter/miscellaneous/canny_edge_detection.py

@@ -3,7 +3,7 @@
 import cv2
 import numpy as np
 
-from nodes.impl.image_utils import normalize, to_uint8
+from nodes.impl.image_utils import to_uint8
 from nodes.properties import expression
 from nodes.properties.inputs import ImageInput, NumberInput
 from nodes.properties.outputs import ImageOutput
@@ -30,5 +30,4 @@ def canny_edge_detection_node(
     t_lower: int,
     t_upper: int,
 ) -> np.ndarray:
-    edges = cv2.Canny(to_uint8(img, normalized=True), t_lower, t_upper)
-    return normalize(edges)
+    return cv2.Canny(to_uint8(img, normalized=True), t_lower, t_upper)

backend/src/packages/chaiNNer_standard/image_filter/miscellaneous/high_pass.py

@@ -47,4 +47,4 @@ def high_pass_filter_node(
     if alpha is not None:
         img = np.dstack((img, alpha))
 
-    return np.clip(img, 0, 1)
+    return img

backend/src/packages/chaiNNer_standard/image_filter/noise/add_noise.py

@@ -58,7 +58,8 @@ class NoiseType(Enum):
                         Input0.channels,
                         match Input2 { NoiseColor::Rgb => 3, NoiseColor::Gray => 1 }
                     )
-                }"""
+                }""",
+            assume_normalized=True,
         )
     ],
 )

backend/src/packages/chaiNNer_standard/image_filter/sharpen/high_boost_sharpen.py

@@ -53,6 +53,4 @@ def sharpen_hbf_node(
         kernel = identity - np.array([[0, 1, 0], [1, 1, 1], [0, 1, 0]]) / 5
 
     kernel = kernel * amount + identity
-    filtered_img = cv2.filter2D(img, -1, kernel)
-
-    return np.clip(filtered_img, 0, 1)
+    return cv2.filter2D(img, -1, kernel)

backend/src/packages/chaiNNer_standard/image_filter/sharpen/sharpen.py

@@ -57,4 +57,4 @@ def sharpen_node(
         diff = np.sign(diff) * np.maximum(0, np.abs(diff) - threshold)
         img += diff * amount
 
-    return np.clip(img, 0, 1)
+    return img

backend/src/packages/chaiNNer_standard/image_utility/compositing/blend.py

@@ -31,7 +31,8 @@
                 width="max(Input0.width, Input1.width)",
                 height="max(Input0.height, Input1.height)",
                 channels="max(Input0.channels, Input1.channels)",
-            )
+            ),
+            assume_normalized=True,
         ),
     ],
 )

backend/src/packages/chaiNNer_standard/image_utility/compositing/caption.py

@@ -30,7 +30,8 @@
                     height: Input0.height + captionHeight,
                     channels: Input0.channels,
                 }
-                """
+                """,
+            assume_normalized=True,
         )
     ],
 )