chore: image_util.assets is a package

Author: keturn

invokeai/app/invocations/composition-nodes.py

@@ -29,10 +29,10 @@
     srgb_from_okhsv,
 )
 from invokeai.backend.image_util.composition import (
-    CIELAB_TO_UPLAB_ICC_PATH,
     MAX_FLOAT,
     equivalent_achromatic_lightness,
     gamut_clip_tensor,
+    get_uplab_profile,
     remove_nans,
     srgb_from_linear_srgb,
     tensor_from_pil_image,
@@ -118,8 +118,7 @@ def invoke(self, context: InvocationContext) -> ImageOutput:
         if self.preserve_lightness or (space == "lch") or (space == "uplab"):
             profile_srgb = ImageCms.createProfile("sRGB")
             if space == "uplab":
-                with open(CIELAB_TO_UPLAB_ICC_PATH, "rb") as f:
-                    profile_uplab = ImageCms.getOpenProfile(f)
+                profile_uplab = get_uplab_profile()
             if profile_uplab is None:
                 profile_lab = ImageCms.createProfile("LAB", colorTemp=6500)
             else:

invokeai/backend/image_util/assets/__init__.py

@@ -7,12 +7,13 @@
 # Copyright (c) 2023 Darren Ringer <dwringer@gmail.com>
 # Parts based on Oklab: Copyright (c) 2021 Bj�rn Ottosson <https://bottosson.github.io/>
 # HSL code based on CPython: Copyright (c) 2001-2023 Python Software Foundation; All Rights Reserved
+import importlib.resources
 from math import pi as PI
-from pathlib import Path
 
 import torch
-from PIL import Image
+from PIL import Image, ImageCms
 
+from invokeai.backend.image_util import assets
 from invokeai.backend.image_util.color_conversion import (
     gamut_clip_tensor,
 )
@@ -23,8 +24,11 @@
 
 MAX_FLOAT = torch.finfo(torch.tensor(1.0).dtype).max
 
+
 # CIE Lab to Uniform Perceptual Lab profile is copyright © 2003 Bruce Justin Lindbloom. All rights reserved. <http://www.brucelindbloom.com>
-CIELAB_TO_UPLAB_ICC_PATH = Path(__file__).parent / "assets" / "CIELab_to_UPLab.icc"
+def get_uplab_profile():
+    with importlib.resources.open_binary(assets, "CIELab_to_UPLab.icc") as icc:
+        return ImageCms.getOpenProfile(icc)
 
 
 def equivalent_achromatic_lightness(lch_tensor: torch.Tensor):