Merge pull request #4740 from awsr/image-type-fix

(minor) PIL Image.Image type fix

cli/gen-styles.py

@@ -24,7 +24,7 @@ options = {
 styles = []
 
 
-def pil_to_b64(img: Image, size: int, quality: int):
+def pil_to_b64(img: Image.Image, size: int, quality: int):
     img = img.convert('RGB')
     img = img.resize((size, size))
     buffer = io.BytesIO()

cli/image-exif.py

@@ -33,7 +33,7 @@ class Exif: # pylint: disable=single-string-used-for-slots
             return self.__dict__[attr]
         return self.exif.get(attr, None)
 
-    def load(self, img: Image):
+    def load(self, img: Image.Image):
         img.load() # exif may not be ready
         exif_dict = {}
         try:

cli/process.py

@@ -35,7 +35,7 @@ class Result():
         self.steps = requested
 
 
-def detect_blur(image: Image):
+def detect_blur(image: Image.Image):
     # based on <https://github.com/karthik9319/Blur-Detection/>
     bw = ImageOps.grayscale(image)
     cx, cy = image.size[0] // 2, image.size[1] // 2
@@ -49,7 +49,7 @@ def detect_blur(image: Image):
     return mean
 
 
-def detect_dynamicrange(image: Image):
+def detect_dynamicrange(image: Image.Image):
     # based on <https://towardsdatascience.com/measuring-enhancing-image-quality-attributes-234b0f250e10>
     data = np.asarray(image)
     image = np.float32(data)
@@ -68,7 +68,7 @@ def detect_dynamicrange(image: Image):
     return round(res, 2)
 
 
-def detect_simmilar(image: Image):
+def detect_simmilar(image: Image.Image):
     img = image.resize((options.process.similarity_size, options.process.similarity_size))
     img = ImageOps.grayscale(img)
     data = np.array(img)

modules/control/proc/marigold/__init__.py

@@ -21,7 +21,7 @@ class MarigoldDetector:
 
     def __call__(
         self,
-        input_image: Image,
+        input_image: Image.Image,
         denoising_steps: int = 10,
         ensemble_size: int = 10,
         processing_res: int = 768,

modules/control/proc/marigold/marigold_pipeline.py

@@ -105,7 +105,7 @@ class MarigoldPipeline(DiffusionPipeline):
     @torch.no_grad()
     def __call__(
         self,
-        input_image: Image,
+        input_image: Image.Image,
         denoising_steps: int = 10,
         ensemble_size: int = 10,
         processing_res: int = 768,

modules/control/processors.py

@@ -304,7 +304,7 @@ class Processor:
             display(e, 'Control Processor load')
             return f'Processor load filed: {processor_id}'
 
-    def __call__(self, image_input: Image, mode: str = 'RGB', width: int = 0, height: int = 0, resize_mode: int = 0, resize_name: str = 'None', scale_tab: int = 1, scale_by: float = 1.0, local_config: dict | None = None):
+    def __call__(self, image_input: Image.Image, mode: str = 'RGB', width: int = 0, height: int = 0, resize_mode: int = 0, resize_name: str = 'None', scale_tab: int = 1, scale_by: float = 1.0, local_config: dict | None = None):
         """Run the preprocessor on an input image and return the processed control map.
 
         Args:

modules/gr_tempdir.py

@@ -49,7 +49,7 @@ def check_tmp_file(gradio, filename):
     return ok
 
 
-def pil_to_temp_file(self, img: Image, dir: str, format="png") -> str: # pylint: disable=redefined-builtin,unused-argument
+def pil_to_temp_file(self, img: Image.Image, dir: str, format="png") -> str: # pylint: disable=redefined-builtin,unused-argument
     """
     # original gradio implementation
     bytes_data = gr.processing_utils.encode_pil_to_bytes(img, format)

modules/masking.py

@@ -257,7 +257,7 @@ def run_segment(input_image: gr.Image, input_mask: np.ndarray):
     return combined_mask
 
 
-def run_rembg(input_image: Image, input_mask: np.ndarray):
+def run_rembg(input_image: Image.Image, input_mask: np.ndarray):
     try:
         import rembg
     except Exception as e:

modules/processing_helpers.py

@@ -106,7 +106,7 @@ def apply_color_correction(correction, original_image, method='histogram'):
     return fn(correction, original_image)
 
 
-def apply_overlay(image: Image, paste_loc, index, overlays):
+def apply_overlay(image: Image.Image, paste_loc, index, overlays):
     if overlays is None or index >= len(overlays):
         return image
     debug(f'Apply overlay: image={image} loc={paste_loc} index={index} overlays={overlays}')

modules/seedvr/src/utils/color_fix.py

@@ -5,7 +5,7 @@ from torch.nn import functional as F
 from ..common.half_precision_fixes import safe_pad_operation, safe_interpolate_operation
 from torchvision.transforms import ToTensor, ToPILImage
 
-def adain_color_fix(target: Image, source: Image):
+def adain_color_fix(target: Image.Image, source: Image.Image):
     # Convert images to tensors
     to_tensor = ToTensor()
     target_tensor = to_tensor(target).unsqueeze(0)
@@ -20,7 +20,7 @@ def adain_color_fix(target: Image, source: Image):
 
     return result_image
 
-def wavelet_color_fix(target: Image, source: Image):
+def wavelet_color_fix(target: Image.Image, source: Image.Image):
     # Convert images to tensors
     to_tensor = ToTensor()
     target_tensor = to_tensor(target).unsqueeze(0)

modules/upscaler_algo.py

@@ -13,7 +13,7 @@ class UpscalerDCC(Upscaler):
             UpscalerData("DCC Interpolation", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         import math
         import numpy as np
         from modules.postprocess.dcc import DCC
@@ -41,7 +41,7 @@ class UpscalerVIPS(Upscaler):
             UpscalerData("VIPS MagicKernelSharp 2021", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         if selected_model is None:
             return img
         from installer import install
@@ -85,7 +85,7 @@ class UpscalerHQX(Upscaler):
             UpscalerData("HQX Interpolation", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         import numpy as np
         from modules.postprocess.hqx import hqx
         t0 = time.time()
@@ -106,7 +106,7 @@ class UpscalerICBI(Upscaler):
             UpscalerData("ICB Interpolation", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         import numpy as np
         from modules.postprocess.icbi import icbi
         t0 = time.time()

modules/upscaler_simple.py

@@ -31,7 +31,7 @@ class UpscalerResize(Upscaler):
             UpscalerData("Resize Sharpfin Lanczos3", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         if selected_model is None:
             return img
         elif selected_model == "Resize Nearest":
@@ -74,7 +74,7 @@ class UpscalerLatent(Upscaler):
             UpscalerData("Latent Bicubic antialias", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         import torch
         import torch.nn.functional as F
         if isinstance(img, torch.Tensor) and (len(img.shape) == 4):

modules/upscaler_spandrel.py

@@ -37,7 +37,7 @@ class UpscalerSpandrel(Upscaler):
         log.debug(f'Upscale: name="{self.selected}" input={img.size} output={upscaled.size} time={t1 - t0:.2f}')
         return upscaled
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         from installer import install
         if selected_model is None:
             return img

modules/upscaler_vae.py

@@ -15,7 +15,7 @@ class UpscalerAsymmetricVAE(Upscaler):
             UpscalerData("Asymmetric VAE v2", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         if selected_model is None:
             return img
         import diffusers
@@ -55,7 +55,7 @@ class UpscalerWanUpscale(Upscaler):
             UpscalerData("WAN Asymmetric Upscale", None, self),
         ]
 
-    def do_upscale(self, img: Image, selected_model=None):
+    def do_upscale(self, img: Image.Image, selected_model=None):
         if selected_model is None:
             return img
         import torch.nn.functional as FN

scripts/daam/trace.py

@@ -34,7 +34,7 @@ class DiffusionHeatMapHooker(AggregateHooker):
         locate_middle = load_heads or save_heads
         self.locator = UNetCrossAttentionLocator(restrict={0} if low_memory else None, locate_middle_block=locate_middle)
         self.last_prompt: str = ''
-        self.last_image: Image = None
+        self.last_image: Image.Image = None
         self.time_idx = 0
         self._gen_idx = 0