resolve merge conflicts

invokeai/backend/stable_diffusion/__init__.py

@@ -1,11 +1,15 @@
 """
 Initialization file for the invokeai.backend.stable_diffusion package
 """
-from .diffusers_pipeline import (
+from .diffusers_pipeline import (  # noqa: F401
     ConditioningData,
     PipelineIntermediateState,
     StableDiffusionGeneratorPipeline,
 )
-from .diffusion import InvokeAIDiffuserComponent
-from .diffusion.cross_attention_map_saving import AttentionMapSaver
-from .diffusion.shared_invokeai_diffusion import PostprocessingSettings, BasicConditioningInfo, SDXLConditioningInfo
+from .diffusion import InvokeAIDiffuserComponent  # noqa: F401
+from .diffusion.cross_attention_map_saving import AttentionMapSaver  # noqa: F401
+from .diffusion.shared_invokeai_diffusion import (  # noqa: F401
+    PostprocessingSettings,
+    BasicConditioningInfo,
+    SDXLConditioningInfo,
+)

invokeai/backend/stable_diffusion/diffusion/__init__.py

@@ -1,9 +1,9 @@
 """
 Initialization file for invokeai.models.diffusion
 """
-from .cross_attention_control import InvokeAICrossAttentionMixin
-from .cross_attention_map_saving import AttentionMapSaver
-from .shared_invokeai_diffusion import (
+from .cross_attention_control import InvokeAICrossAttentionMixin  # noqa: F401
+from .cross_attention_map_saving import AttentionMapSaver  # noqa: F401
+from .shared_invokeai_diffusion import (  # noqa: F401
     InvokeAIDiffuserComponent,
     PostprocessingSettings,
     BasicConditioningInfo,

invokeai/backend/stable_diffusion/diffusion/cross_attention_control.py

@@ -4,6 +4,7 @@
 
 import enum
 import math
+from dataclasses import dataclass, field
 from typing import Callable, Optional
 
 import diffusers
@@ -12,6 +13,11 @@ import torch
 from compel.cross_attention_control import Arguments
 from diffusers.models.unet_2d_condition import UNet2DConditionModel
 from diffusers.models.attention_processor import AttentionProcessor
+from diffusers.models.attention_processor import (
+    Attention,
+    AttnProcessor,
+    SlicedAttnProcessor,
+)
 from torch import nn
 
 import invokeai.backend.util.logging as logger
@@ -522,14 +528,6 @@ class AttnProcessor:
         return hidden_states
 
 """
-from dataclasses import dataclass, field
-
-import torch
-from diffusers.models.attention_processor import (
-    Attention,
-    AttnProcessor,
-    SlicedAttnProcessor,
-)
 
 
 @dataclass

invokeai/backend/stable_diffusion/diffusion/cross_attention_map_saving.py

@@ -5,8 +5,6 @@ import torch
 from torchvision.transforms.functional import InterpolationMode
 from torchvision.transforms.functional import resize as tv_resize
 
-from .cross_attention_control import CrossAttentionType, get_cross_attention_modules
-
 
 class AttentionMapSaver:
     def __init__(self, token_ids: range, latents_shape: torch.Size):

invokeai/backend/stable_diffusion/diffusion/shared_invokeai_diffusion.py

@@ -3,15 +3,12 @@ from __future__ import annotations
 from contextlib import contextmanager
 from dataclasses import dataclass
 import math
-from typing import Any, Callable, Dict, Optional, Union, List
+from typing import Any, Callable, Optional, Union
 
-import numpy as np
 import torch
 from diffusers import UNet2DConditionModel
-from diffusers.models.attention_processor import AttentionProcessor
 from typing_extensions import TypeAlias
 
-import invokeai.backend.util.logging as logger
 from invokeai.app.services.config import InvokeAIAppConfig
 
 from .cross_attention_control import (
@@ -579,7 +576,7 @@ class InvokeAIDiffuserComponent:
         latents.to(device="cpu")
 
         if (
-            h_symmetry_time_pct != None
+            h_symmetry_time_pct is not None
             and self.last_percent_through < h_symmetry_time_pct
             and percent_through >= h_symmetry_time_pct
         ):
@@ -595,7 +592,7 @@ class InvokeAIDiffuserComponent:
             )
 
         if (
-            v_symmetry_time_pct != None
+            v_symmetry_time_pct is not None
             and self.last_percent_through < v_symmetry_time_pct
             and percent_through >= v_symmetry_time_pct
         ):

invokeai/backend/stable_diffusion/image_degradation/__init__.py

@@ -1,6 +1,6 @@
-from ldm.modules.image_degradation.bsrgan import (
+from ldm.modules.image_degradation.bsrgan import (  # noqa: F401
     degradation_bsrgan_variant as degradation_fn_bsr,
 )
-from ldm.modules.image_degradation.bsrgan_light import (
+from ldm.modules.image_degradation.bsrgan_light import (  # noqa: F401
     degradation_bsrgan_variant as degradation_fn_bsr_light,
 )

invokeai/backend/stable_diffusion/image_degradation/bsrgan.py

@@ -573,14 +573,15 @@ def degradation_bsrgan_variant(image, sf=4, isp_model=None):
     hq: corresponding high-quality patch, size: (lq_patchsizexsf)X(lq_patchsizexsf)XC, range: [0, 1]
     """
     image = util.uint2single(image)
-    isp_prob, jpeg_prob, scale2_prob = 0.25, 0.9, 0.25
-    sf_ori = sf
+    jpeg_prob, scale2_prob = 0.9, 0.25
+    # isp_prob = 0.25  # uncomment with `if i== 6` block below
+    # sf_ori = sf  # uncomment with `if i== 6` block below
 
     h1, w1 = image.shape[:2]
     image = image.copy()[: w1 - w1 % sf, : h1 - h1 % sf, ...]  # mod crop
     h, w = image.shape[:2]
 
-    hq = image.copy()
+    # hq = image.copy()  # uncomment with `if i== 6` block below
 
     if sf == 4 and random.random() < scale2_prob:  # downsample1
         if np.random.rand() < 0.5:
@@ -777,7 +778,7 @@ if __name__ == "__main__":
         img_lq_bicubic = albumentations.SmallestMaxSize(max_size=h, interpolation=cv2.INTER_CUBIC)(image=img)["image"]
         print(img_lq.shape)
         print("bicubic", img_lq_bicubic.shape)
-        print(img_hq.shape)
+        # print(img_hq.shape)
         lq_nearest = cv2.resize(
             util.single2uint(img_lq),
             (int(sf * img_lq.shape[1]), int(sf * img_lq.shape[0])),
@@ -788,5 +789,6 @@ if __name__ == "__main__":
             (int(sf * img_lq.shape[1]), int(sf * img_lq.shape[0])),
             interpolation=0,
         )
-        img_concat = np.concatenate([lq_bicubic_nearest, lq_nearest, util.single2uint(img_hq)], axis=1)
+        # img_concat = np.concatenate([lq_bicubic_nearest, lq_nearest, util.single2uint(img_hq)], axis=1)
+        img_concat = np.concatenate([lq_bicubic_nearest, lq_nearest], axis=1)
         util.imsave(img_concat, str(i) + ".png")

invokeai/backend/stable_diffusion/image_degradation/bsrgan_light.py

@@ -577,14 +577,15 @@ def degradation_bsrgan_variant(image, sf=4, isp_model=None):
     hq: corresponding high-quality patch, size: (lq_patchsizexsf)X(lq_patchsizexsf)XC, range: [0, 1]
     """
     image = util.uint2single(image)
-    isp_prob, jpeg_prob, scale2_prob = 0.25, 0.9, 0.25
-    sf_ori = sf
+    jpeg_prob, scale2_prob = 0.9, 0.25
+    # isp_prob = 0.25  # uncomment with `if i== 6` block below
+    # sf_ori = sf  # uncomment with `if i== 6` block below
 
     h1, w1 = image.shape[:2]
     image = image.copy()[: w1 - w1 % sf, : h1 - h1 % sf, ...]  # mod crop
     h, w = image.shape[:2]
 
-    hq = image.copy()
+    # hq = image.copy()  # uncomment with `if i== 6` block below
 
     if sf == 4 and random.random() < scale2_prob:  # downsample1
         if np.random.rand() < 0.5:

invokeai/backend/stable_diffusion/image_degradation/utils_image.py

@@ -8,8 +8,6 @@ import numpy as np
 import torch
 from torchvision.utils import make_grid
 
-# import matplotlib.pyplot as plt   # TODO: check with Dominik, also bsrgan.py vs bsrgan_light.py
-
 import invokeai.backend.util.logging as logger
 
 os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
@@ -50,6 +48,8 @@ def get_timestamp():
 
 
 def imshow(x, title=None, cbar=False, figsize=None):
+    import matplotlib.pyplot as plt
+
     plt.figure(figsize=figsize)
     plt.imshow(np.squeeze(x), interpolation="nearest", cmap="gray")
     if title:
@@ -60,6 +60,8 @@ def imshow(x, title=None, cbar=False, figsize=None):
 
 
 def surf(Z, cmap="rainbow", figsize=None):
+    import matplotlib.pyplot as plt
+
     plt.figure(figsize=figsize)
     ax3 = plt.axes(projection="3d")
 

invokeai/backend/stable_diffusion/schedulers/__init__.py

@@ -1 +1 @@
-from .schedulers import SCHEDULER_MAP
+from .schedulers import SCHEDULER_MAP  # noqa: F401