nodes/Shibiko-AI-ShibikoAI-ComfyUI-Tools
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Tensor Batch Fixes

Browse Source
This commit is contained in:
Brian Jemilo II
2025-04-12 22:51:14 -05:00
parent b353b48264
commit 92242f43b1
15 changed files with 866 additions and 282 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CONTRIBUTORS.md
+44
View File
@@ -0,0 +1,44 @@
# Contributors and Attributions
This project integrates and builds upon work from the following developers and repositories.
---
### Third-Party Comfy UI Nodes
- **WD14-Tagger**
- **Author:** pythongosssss
- **Source:** https://github.com/pythongosssss/stable-diffusion-webui-wd14-tagger
- **License:** MIT
- **Usage:**
- `WD14Tagger`: Tagging images with anime-style concept labels
- **Note:** Switched from my own WD14-Tagger implementation due to its extra convenient features.
- **Essentials**
- **Author:** Matteo Spinelli
- **Source:** https://github.com/mat-sz/ComfyUI-Utilities
- **License:** MIT
- **Usage:**
- `ImageApplyLUT`: Applying LUTs to images
- **Note:** I adopted this implementation over my own original node due to better customization.
- **WAS NODE Suite**
- **Author:** Jordan Thompson (WASasquatch)
- **Source:** https://github.com/WASasquatch/was-node-suite-comfyui
- **License:** MIT
- **Usage:**
- `WAS_Image_Filters`: Used for advanced image filter processing
- `WAS_Remove_Rembg`: Used for removing backgrounds efficiently
- **Note:** Replaced my earlier attempts with WAS nodes because I preferred their feature set and reliability.
---
### Special Thanks
- These developers significantly improved the quality and capability of this project through their publicly shared work.
---
### Original Work
All other code and systems are original to this repository’s authors.
LICENSE
+17 -197
View File
@@ -1,201 +1,21 @@
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
MIT License
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
Copyright (c) 2025 Brian Jemilo II
1. Definitions.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright 2024 Brian Jemilo II
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
__init__.py
+17 -18
View File
@@ -1,28 +1,27 @@
from .nodes.cascades import NODE_CLASS_MAPPINGS as CASCADES_NODE_CLASS_MAPPINGS
from .nodes.cascades import NODE_DISPLAY_NAME_MAPPINGS as CASCADES_NODE_DISPLAY_NAME_MAPPINGS
from .nodes.luts import NODE_CLASS_MAPPINGS as LUTS_NODE_CLASS_MAPPINGS
from .nodes.luts import NODE_DISPLAY_NAME_MAPPINGS as LUTS_NODE_DISPLAY_NAME_MAPPINGS
from .nodes.waifu2x import NODE_CLASS_MAPPINGS as WAIFU2X_NODE_CLASS_MAPPINGS
from .nodes.waifu2x import NODE_DISPLAY_NAME_MAPPINGS as WAIFU2X_NODE_DISPLAY_NAME_MAPPINGS
import os
import importlib
WEB_DIRECTORY = "./web"
NODE_CLASS_MAPPINGS = {}
NODE_DISPLAY_NAME_MAPPINGS = {}
# Cascades
NODE_CLASS_MAPPINGS.update(CASCADES_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(CASCADES_NODE_DISPLAY_NAME_MAPPINGS)
print(f'\033[93m[Shibiko AI] \033[31mCascades \033[92mLoaded\033[0m')
# Luts
NODE_CLASS_MAPPINGS.update(LUTS_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(LUTS_NODE_DISPLAY_NAME_MAPPINGS)
print(f'\033[93m[Shibiko AI] \033[31mLuts \033[92mLoaded\033[0m')
def load_module(module_name):
try:
module = importlib.import_module(f'.nodes.{module_name}', package=__package__)
NODE_CLASS_MAPPINGS.update(getattr(module, 'NODE_CLASS_MAPPINGS', {}))
NODE_DISPLAY_NAME_MAPPINGS.update(getattr(module, 'NODE_DISPLAY_NAME_MAPPINGS', {}))
print(f'\033[93m[Shibiko AI] \033[31m{module_name.capitalize()} \033[92mloaded\033[0m')
except ImportError as e:
print(f'\033[93m[Shibiko AI] \033[31mFailed to load {module_name}: {str(e)}\033[0m')
# Waifu2x
NODE_CLASS_MAPPINGS.update(WAIFU2X_NODE_CLASS_MAPPINGS)
NODE_DISPLAY_NAME_MAPPINGS.update(WAIFU2X_NODE_DISPLAY_NAME_MAPPINGS)
print(f'\033[93m[Shibiko AI] \033[31mWaifu2x \033[92mLoaded\033[0m')
current_dir = os.path.dirname(os.path.abspath(__file__))
nodes_dir = os.path.join(current_dir, 'nodes')
for filename in os.listdir(nodes_dir):
if filename.endswith('.py') and not filename.startswith('__'):
load_module(filename[:-3])
__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY']
print(f'\033[93m[Shibiko AI] \033[92mLoading Complete!\033[0m')
nodes/bbox_insert_image.py
+91
View File
@@ -0,0 +1,91 @@
import numpy as np
import torch
import torch.nn.functional as F
class BboxInsertImage:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"image": ("IMAGE",),
"bbox_image": ("IMAGE",),
"bbox": ("BBOX",),
"blend_factor": ("FLOAT", {"default": 0.5, "min": 0.0, "max": 1.0, "step": 0.1}),
}
}
CATEGORY = "Shibiko AI"
DESCRIPTION = "Insert Image into Bounding Box Area with Smooth Blending."
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
FUNCTION = "__call__"
def __call__(self, image, bbox_image, bbox, blend_factor=0.5):
device = image.device
bbox_image = bbox_image.to(device)
if isinstance(bbox, (np.ndarray, torch.Tensor)):
bbox = bbox.tolist()
if len(bbox) == 1 and isinstance(bbox[0], (list, tuple)):
bbox = bbox[0]
if len(bbox) != 4:
raise ValueError(f"Invalid bbox format. Expected 4 values, but got {len(bbox)}")
x, y, w, h = map(int, bbox)
# Ensure bbox_image has 4 dimensions (N, H, W, C)
if bbox_image.dim() == 3:
bbox_image = bbox_image.unsqueeze(0)
# If bbox_image doesn't have a channel dimension, add it
if bbox_image.shape[-1] != image.shape[-1]:
bbox_image = bbox_image.unsqueeze(-1).repeat(1, 1, 1, image.shape[-1])
# Get the dimensions of the original region
original_region = image[:, y:y + h, x:x + w, :]
original_h, original_w = original_region.shape[1:3]
# Resize bbox_image to match the original region size
bbox_image = F.interpolate(bbox_image.permute(0, 3, 1, 2),
size=(original_h, original_w),
mode='bilinear',
align_corners=False)
bbox_image = bbox_image.permute(0, 2, 3, 1)
# Ensure bbox_image has the same batch size as image
if bbox_image.shape[0] != image.shape[0]:
bbox_image = bbox_image.repeat(image.shape[0], 1, 1, 1)
# Create a Gaussian blending mask
blend_mask = self.create_gaussian_mask(original_h, original_w, device)
# Apply blending
blended_image = blend_factor * bbox_image + (1 - blend_factor) * original_region
blended_image = blend_mask * bbox_image + (1 - blend_mask) * original_region
# Insert blended image into original image
image[:, y:y + h, x:x + w, :] = blended_image
return (image,)
def create_gaussian_mask(self, height, width, device, sigma=10):
"""
Creates a Gaussian blend mask that smoothly fades from the center outwards.
"""
y, x = torch.meshgrid(torch.linspace(-1, 1, height, device=device),
torch.linspace(-1, 1, width, device=device), indexing='ij')
distance = torch.sqrt(x ** 2 + y ** 2)
mask = torch.exp(-distance ** 2 / (2 * sigma ** 2))
mask = (mask - mask.min()) / (mask.max() - mask.min()) # Normalize to 0-1
return mask.unsqueeze(0).unsqueeze(-1) # Add batch and channel dims
NODE_CLASS_MAPPINGS = {"BboxInsertImage": BboxInsertImage}
NODE_DISPLAY_NAME_MAPPINGS = {"BboxInsertImage": "Shibiko AI - BBOX Insert Image"}
nodes/bbox_split.py
+24
View File
@@ -0,0 +1,24 @@
class BboxSplit:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"bbox": ("BBOX",),
}
}
CATEGORY = "Shibiko AI"
DESCRIPTION = "Split a BBOX into x, y, width, height."
RETURN_TYPES = ("INT", "INT", "INT", "INT")
RETURN_NAMES = ("x", "y", "width", "height")
FUNCTION = "__call__"
def __call__(self, bbox):
x, y, width, height = bbox[0]
return x, y, width, height,
NODE_CLASS_MAPPINGS = {"BboxSplit": BboxSplit}
NODE_DISPLAY_NAME_MAPPINGS = {"BboxSplit": "Shibiko AI - BBOX Split"}
nodes/cascades.py
+4 -3
View File
@@ -31,7 +31,7 @@ class Cascade:
},
}
CATEGORY = "Shibiko"
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE", "MASK", "BBOX")
RETURN_NAMES = ("image", "mask", "bbox")
@@ -71,7 +71,7 @@ class Cascade:
gray = self.image
# Detect faces in the image
self.bbox = self.haar_cascade_face.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(24, 24))
self.bbox = self.haar_cascade_face.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=5, minSize=(32, 32))
return self.bbox
def draw(self, image=None, padding=None):
@@ -101,7 +101,7 @@ class Cascade:
for (x, y, w, h) in self.bbox:
mask[y - self.padding:y + h + self.padding, x - self.padding:x + w + self.padding] = 255
if blur is not None:
if blur is not None and blur > 0:
if blur_type == 'gaussian':
blur = blur if blur % 2 == 1 else blur + 1
dilation = dilation if dilation % 2 == 1 else dilation + 1
@@ -145,6 +145,7 @@ class Cascade:
cascade='frontalface_default',
dilation=4,
padding=50,
**kwargs
):
if cascade != self.cascade:
self.load(cascade)
nodes/image_filters.py
+226
View File
@@ -0,0 +1,226 @@
import numpy as np
import torch
from PIL import Image, ImageEnhance, ImageFilter
from ..utils.convert import tensor2pil, pil2tensor
# From the original code of WAS Node Suite with some additions and modifications
class ImageFilters:
def __init__(self):
pass
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"image": ("IMAGE",),
"brightness": ("FLOAT", {"default": 0.0, "min": -1.0, "max": 1.0, "step": 0.01}),
"contrast": ("FLOAT", {"default": 1.0, "min": -1.0, "max": 2.0, "step": 0.01}),
"saturation": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 5.0, "step": 0.01}),
"sharpness": ("FLOAT", {"default": 1.0, "min": -5.0, "max": 5.0, "step": 0.01}),
"blur": ("INT", {"default": 0, "min": 0, "max": 16, "step": 1}),
"sharpen": ("INT", {"default": 0, "min": 0, "max": 16, "step": 1}),
"smooth": ("INT", {"default": 0, "min": 0, "max": 16, "step": 1}),
"smooth_more": ("INT", {"default": 0, "min": 0, "max": 16, "step": 1}),
"gaussian_blur": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1024.0, "step": 0.1}),
"edge_enhance": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.01}),
"detail_enhance": (["false", "true"],),
},
}
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
FUNCTION = "__call__"
def __call__(self,
image,
brightness=0.0,
contrast=1.0,
saturation=1.0,
sharpness=1.0,
blur=0,
sharpen=0,
smooth=0,
smooth_more=0,
gaussian_blur=0.0,
edge_enhance=0.0,
detail_enhance=False,
**kwargs):
tensors = []
if len(image) > 1:
for img in image:
pil_image = None
# Apply NP Adjustments
if brightness > 0.0 or brightness < 0.0:
# Apply brightness
img = np.clip(img + brightness, 0.0, 1.0)
if contrast > 1.0 or contrast < 1.0:
# Apply contrast
img = np.clip(img * contrast, 0.0, 1.0)
# Apply PIL Adjustments
if saturation > 1.0 or saturation < 1.0:
# PIL Image
pil_image = tensor2pil(img)
# Apply saturation
pil_image = ImageEnhance.Color(pil_image).enhance(saturation)
if sharpness > 1.0 or sharpness < 1.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply sharpness
pil_image = ImageEnhance.Sharpness(pil_image).enhance(sharpness)
if blur > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(blur):
pil_image = pil_image.filter(ImageFilter.BLUR)
if sharpen > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(sharpen):
pil_image = pil_image.filter(ImageFilter.SHARPEN)
if smooth > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(smooth):
pil_image = pil_image.filter(ImageFilter.SMOOTH)
if smooth_more > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(smooth_more):
pil_image = pil_image.filter(ImageFilter.SMOOTH_MORE)
if gaussian_blur > 0.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply Gaussian blur
pil_image = pil_image.filter(ImageFilter.GaussianBlur(radius=gaussian_blur))
if edge_enhance > 0.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Edge Enhancement
edge_enhanced_img = pil_image.filter(ImageFilter.EDGE_ENHANCE_MORE)
# Blend Mask
blend_mask = Image.new(mode="L", size=pil_image.size, color=(round(edge_enhance * 255)))
# Composite Original and Enhanced Version
pil_image = Image.composite(edge_enhanced_img, pil_image, blend_mask)
# Clean-up
del blend_mask, edge_enhanced_img
if detail_enhance == "true" or detail_enhance is True:
pil_image = pil_image if pil_image else tensor2pil(img)
pil_image = pil_image.filter(ImageFilter.DETAIL)
# Output image
out_image = (pil2tensor(pil_image) if pil_image else img.unsqueeze(0))
tensors.append(out_image)
tensors = torch.cat(tensors, dim=0)
else:
pil_image = None
img = image
# Apply NP Adjustments
if brightness > 0.0 or brightness < 0.0:
# Apply brightness
img = np.clip(img + brightness, 0.0, 1.0)
if contrast > 1.0 or contrast < 1.0:
# Apply contrast
img = np.clip(img * contrast, 0.0, 1.0)
# Apply PIL Adjustments
if saturation > 1.0 or saturation < 1.0:
# PIL Image
pil_image = tensor2pil(img)
# Apply saturation
pil_image = ImageEnhance.Color(pil_image).enhance(saturation)
if sharpness > 1.0 or sharpness < 1.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply sharpness
pil_image = ImageEnhance.Sharpness(pil_image).enhance(sharpness)
if blur > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(blur):
pil_image = pil_image.filter(ImageFilter.BLUR)
if sharpen > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(sharpen):
pil_image = pil_image.filter(ImageFilter.SHARPEN)
if smooth > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(smooth):
pil_image = pil_image.filter(ImageFilter.SMOOTH)
if smooth_more > 0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply blur
for _ in range(smooth_more):
pil_image = pil_image.filter(ImageFilter.SMOOTH_MORE)
if gaussian_blur > 0.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Apply Gaussian blur
pil_image = pil_image.filter(
ImageFilter.GaussianBlur(radius=gaussian_blur))
if edge_enhance > 0.0:
# Assign or create PIL Image
pil_image = pil_image if pil_image else tensor2pil(img)
# Edge Enhancement
edge_enhanced_img = pil_image.filter(ImageFilter.EDGE_ENHANCE_MORE)
# Blend Mask
blend_mask = Image.new(
mode="L", size=pil_image.size, color=(round(edge_enhance * 255)))
# Composite Original and Enhanced Version
pil_image = Image.composite(
edge_enhanced_img, pil_image, blend_mask)
# Clean-up
del blend_mask, edge_enhanced_img
if detail_enhance == "true":
pil_image = pil_image if pil_image else tensor2pil(img)
pil_image = pil_image.filter(ImageFilter.DETAIL)
# Output image
out_image = (pil2tensor(pil_image) if pil_image else img)
tensors = out_image
return (tensors, )
NODE_CLASS_MAPPINGS = {"ImageFilters": ImageFilters}
NODE_DISPLAY_NAME_MAPPINGS = {"Cascade": "Shibiko AI - Image Filters"}
nodes/luts.py
+136 -25
View File
@@ -1,16 +1,22 @@
import os
import folder_paths
import numpy as np
import torch
from pathlib import Path
from PIL import ImageFilter
from colour.io.luts.iridas_cube import read_LUT_IridasCube
from typing import List, Union
from ..utils.convert import convert
from ..utils.convert import convert, tensor2pil, pil2tensor
from ..utils.directory import initialize_directory
class Luts:
luts_directory = initialize_directory('luts')
@classmethod
def INPUT_TYPES(self):
luts = [lut.split(".")[0] for lut in os.listdir(self.luts_directory) if lut.endswith(".cube")]
def INPUT_TYPES(cls):
luts = [lut.split(".")[0] for lut in os.listdir(cls.luts_directory) if lut.endswith(".cube")]
return {
"required": {
@@ -19,7 +25,7 @@ class Luts:
},
}
CATEGORY = "Shibiko"
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
@@ -27,18 +33,19 @@ class Luts:
FUNCTION = "__call__"
def apply_lut(self, image, lut):
print("Applying LUT...")
if not lut.endswith(".cube"):
lut += ".cube"
lut_file_name = lut
lut_directory = initialize_directory('luts')
lut_path = os.path.join(lut_directory, f"{lut_file_name}")
lut = self.read_lut(lut_path)
lut_path = os.path.join(lut_directory, lut_file_name)
lut_filter = self.read_lut(lut_path)
if isinstance(image, list):
image = [img.filter(lut) for img in image]
else:
image = image.filter(lut)
# Convert tensor to PIL, apply LUT, then convert back to tensor
if isinstance(image, torch.Tensor):
image = tensor2pil(image)
# Apply the LUT filter
image = image.filter(lut_filter)
return image
@@ -49,35 +56,139 @@ class Luts:
size = round(len(lut_raw) ** (1 / 3))
lut_table = [
self.row2val(row.split(" ")) for row in lut_raw if self.is_3dlut_row(row.split(" "))
self.row2val(row.split(" ")) for row in lut_raw if self.is_3D_lut_row(row.split(" "))
]
return ImageFilter.Color3DLUT(size, lut_table, num_channels)
@staticmethod
def is_3dlut_row(row: List) -> bool:
def is_3D_lut_row(row: List) -> bool:
"""Check if one line in the file has exactly 3 values"""
row_values = []
for val in row:
try:
row_values.append(float(val))
except:
except ValueError:
return False
if len(row_values) == 3:
return True
return False
return len(row_values) == 3
@staticmethod
def row2val(row):
return tuple([float(val) for val in row])
def __call__(self, image, lut, unique_id):
image = convert(image)
image = self.apply_lut(image, lut)
image = convert(image)
def __call__(self, image, lut, **kwargs):
# Handle single image or batch
if image.dim() == 3: # Single image (H, W, C)
images = [image]
else: # Batch of images (B, H, W, C)
images = [img for img in image]
return (image,)
processed_images = []
for img in images:
# Convert to PIL
img_pil = tensor2pil(img)
# Apply LUT
img_processed = self.apply_lut(img_pil, lut)
# Convert back to tensor
img_tensor = pil2tensor(img_processed)
processed_images.append(img_tensor)
# Stack all tensors if we have multiple images
if len(processed_images) > 1:
output = torch.cat(processed_images, dim=0)
else:
output = processed_images[0]
# Ensure proper output format (B, H, W, C) with float32 dtype
if output.dtype != torch.float32:
output = output.float()
if output.max() > 1.0:
output = output / 255.0
return (output,)
NODE_CLASS_MAPPINGS = {"Luts": Luts}
NODE_DISPLAY_NAME_MAPPINGS = {"Luts": "Shibiko AI - Luts"}
# From https://github.com/yoonsikp/pycubelut/blob/master/pycubelut.py (MIT license)
class LutsAdvanced:
@classmethod
def INPUT_TYPES(s):
return {
"required": {
"image": ("IMAGE",),
"lut_file": (folder_paths.get_filename_list("luts"), {"default": "Cinematic.cube"}),
"gamma_correction": ("BOOLEAN", {"default": True}),
"clip_values": ("BOOLEAN", {"default": True}),
"strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.1}),
}}
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
FUNCTION = "__call__"
# TODO: check if we can do without numpy
def __call__(self, image, lut_file="cinematic", gamma_correction=True, clip_values=True, strength=1.0):
lut_file_path = folder_paths.get_full_path("luts", lut_file + '.cube')
if not lut_file_path or not Path(lut_file_path).exists():
print(f"Could not find LUT file: {lut_file_path}")
return (image,)
device = image.device
lut = read_LUT_IridasCube(lut_file_path)
lut.name = lut_file
if clip_values:
if lut.domain[0].max() == lut.domain[0].min() and lut.domain[1].max() == lut.domain[1].min():
lut.table = np.clip(lut.table, lut.domain[0, 0], lut.domain[1, 0])
else:
for dim in range(3):
domain0 = lut.domain[0, dim]
domain1 = lut.domain[1, dim]
if len(lut.table.shape) == 2:
lut.table[:, dim] = np.clip(lut.table[:, dim], domain0, domain1)
else: # 3D
lut.table[:, :, :, dim] = np.clip(lut.table[:, :, :, dim], domain0, domain1)
out = []
for img in image: # TODO: is this more resource efficient? should we use a batch instead?
lut_img = img.cpu().numpy().copy()
is_non_default_domain = not np.array_equal(lut.domain, np.array([[0., 0., 0.], [1., 1., 1.]]))
dom_scale = None
if is_non_default_domain:
dom_scale = lut.domain[1] - lut.domain[0]
lut_img = lut_img * dom_scale + lut.domain[0]
if gamma_correction:
lut_img = lut_img ** (1 / 2.2)
lut_img = lut.apply(lut_img)
if gamma_correction:
lut_img = lut_img ** 2.2
if is_non_default_domain:
lut_img = (lut_img - lut.domain[0]) / dom_scale
lut_img = torch.from_numpy(lut_img).to(device)
if strength < 1.0:
lut_img = strength * lut_img + (1 - strength) * img
out.append(lut_img)
out = torch.stack(out)
return (out,)
NODE_CLASS_MAPPINGS = {
"Luts": Luts,
"LutsAdvanced": LutsAdvanced
}
NODE_DISPLAY_NAME_MAPPINGS = {
"Luts": "Shibiko AI - Luts",
"LutsAdvanced": "Shibiko AI - Luts (Advanced)"
}
nodes/remove_area_by_mask.py
+60
View File
@@ -0,0 +1,60 @@
import cv2
import numpy as np
import torch
class RemoveAreaByMask:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"image": ("IMAGE",),
"mask": ("MASK",),
},
"optional": {
"invert_mask": ("BOOLEAN", {"default": False}),
},
}
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
FUNCTION = "__call__"
@staticmethod
def NHWC_to_CV2(image):
image_numpy = image.cpu().numpy()
return (image_numpy[0] * 255).astype(np.uint8) # Convert to uint8, keep NHWC
@staticmethod
def CV2_to_NHWC(image):
image_rgb = image.astype(np.float32) / 255.0
return torch.from_numpy(image_rgb).unsqueeze(0) # (1, H, W, C)
def __call__(self, image, mask, invert_mask=False):
image_cv2 = self.NHWC_to_CV2(image)
mask_cv2 = self.NHWC_to_CV2(mask)
if mask_cv2.ndim == 3 and mask_cv2.shape[2] == 3:
mask_cv2 = cv2.cvtColor(mask_cv2, cv2.COLOR_BGR2GRAY)
if mask_cv2.shape[:2] != image_cv2.shape[:2]:
mask_cv2 = cv2.resize(mask_cv2, (image_cv2.shape[1], image_cv2.shape[0]), interpolation=cv2.INTER_NEAREST)
mask_float = mask_cv2.astype(np.float32) / 255.0
if mask_float.ndim == 2:
mask_float = np.expand_dims(mask_float, axis=-1)
if invert_mask:
mask_float = 1.0 - mask_float
masked_image = (image_cv2 * (1 - mask_float)).astype(np.uint8)
return (self.CV2_to_NHWC(masked_image),) # Return (1, H, W, C)
NODE_CLASS_MAPPINGS = {"RemoveAreaByMask": RemoveAreaByMask}
NODE_DISPLAY_NAME_MAPPINGS = {"RemoveAreaByMask": "Shibiko AI - Remove Area By Mask"}
nodes/remove_background.py
+130
View File
@@ -0,0 +1,130 @@
import os
import torch
import folder_paths
from rembg import remove, new_session
from ..utils.convert import pil2tensor, tensor2pil
class RemoveBackground:
def __init__(self):
pass
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"images": ("IMAGE",),
"transparency": ("BOOLEAN", {"default": True},),
"model": (["u2net", "u2netp", "u2net_human_seg", "silueta", "isnet-general-use", "isnet-anime"], {"default": "isnet-anime"}),
"post_processing": ("BOOLEAN", {"default": False}),
"only_mask": ("BOOLEAN", {"default": False},),
"alpha_matting": ("BOOLEAN", {"default": False},),
"alpha_matting_foreground_threshold": ("INT", {"default": 240, "min": 0, "max": 255}),
"alpha_matting_background_threshold": ("INT", {"default": 10, "min": 0, "max": 255}),
"alpha_matting_erode_size": ("INT", {"default": 10, "min": 0, "max": 255}),
"background_color": (["none", "black", "white", "magenta", "chroma green", "chroma blue"],),
"putalpha": ("BOOLEAN", {"default": False},),
},
}
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
FUNCTION = "__call__"
# A helper function to convert from strings to logical boolean
# Conforms to https://docs.python.org/3/library/stdtypes.html#truth-value-testing
# With the addition of evaluating string representations of Falsy types
@staticmethod
def __convertToBool(x):
# Evaluate string representation of False types
if type(x) is str:
x = x.strip()
if (x.lower() == 'false'
or x.lower() == 'none'
or x == '0'
or x == '0.0'
or x == '0j'
or x == "''"
or x == '""'
or x == "()"
or x == "[]"
or x == "{}"
or x.lower() == "decimal(0)"
or x.lower() == "fraction(0,1)"
or x.lower() == "set()"
or x.lower() == "range(0)"):
return False
else:
return True
return bool(x)
def __call__(
self,
images,
transparency=True,
model="isnet-anime",
alpha_matting=False,
alpha_matting_foreground_threshold=240,
alpha_matting_background_threshold=10,
alpha_matting_erode_size=10,
post_processing=False,
only_mask=False,
background_color="none",
putalpha=False,
**kwargs
):
# ComfyUI will allow strings in place of booleans, validate the input.
transparency = transparency if type(transparency) is bool else self.__convertToBool(transparency)
alpha_matting = alpha_matting if type(alpha_matting) is bool else self.__convertToBool(alpha_matting)
post_processing = post_processing if type(post_processing) is bool else self.__convertToBool(post_processing)
only_mask = only_mask if type(only_mask) is bool else self.__convertToBool(only_mask)
os.environ['U2NET_HOME'] = os.path.join(folder_paths.models_dir, 'rembg')
os.makedirs(os.environ['U2NET_HOME'], exist_ok=True)
# Set background color
if background_color == "black":
bgrgba = [0, 0, 0, 255]
elif background_color == "white":
bgrgba = [255, 255, 255, 255]
elif background_color == "magenta":
bgrgba = [255, 0, 255, 255]
elif background_color == "chroma green":
bgrgba = [0, 177, 64, 255]
elif background_color == "chroma blue":
bgrgba = [0, 71, 187, 255]
else:
bgrgba = None
if transparency and bgrgba is not None:
bgrgba[3] = 0
batch_tensor = []
for image in images:
image = tensor2pil(image)
batch_tensor.append(pil2tensor(
remove(
image,
session=new_session(model),
post_process_mask=post_processing,
alpha_matting=alpha_matting,
alpha_matting_foreground_threshold=alpha_matting_foreground_threshold,
alpha_matting_background_threshold=alpha_matting_background_threshold,
alpha_matting_erode_size=alpha_matting_erode_size,
only_mask=only_mask,
bgcolor=bgrgba,
putalpha=putalpha,
)
.convert(('RGBA' if transparency else 'RGB'))))
batch_tensor = torch.cat(batch_tensor, dim=0)
return (batch_tensor,)
NODE_CLASS_MAPPINGS = {"RemoveBackground": RemoveBackground}
NODE_DISPLAY_NAME_MAPPINGS = {"RemoveBackground": "Shibiko AI - Remove Background"}
nodes/seed.py
+27
View File
@@ -0,0 +1,27 @@
class SeedGenerator:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"seed": ("INT", {"default": 0, "min": 0, "max": 18446744073709551615}),
}
}
@classmethod
def IS_CHANGED(cls, seed=0):
return seed
CATEGORY = "Shibiko AI"
DESCRIPTION = "Share Your Seed Across Multiple Nodes."
RETURN_TYPES = ("INT",)
RETURN_NAMES = ("SEED",)
FUNCTION = "__call__"
def __call__(self, seed=0):
return (seed,)
NODE_CLASS_MAPPINGS = {"SeedGenerator": SeedGenerator}
NODE_DISPLAY_NAME_MAPPINGS = {"SeedGenerator": "Shibiko AI - Seed Generator"}
nodes/waifu2x.py
+54 -39
View File
@@ -1,6 +1,6 @@
import torch
from typing import Optional
from ..utils.convert import convert
from ..utils.convert import convert, tensor2pil, pil2tensor
class Waifu2x:
@@ -17,7 +17,7 @@ class Waifu2x:
},
}
CATEGORY = "Shibiko"
CATEGORY = "Shibiko AI"
RETURN_TYPES = ("IMAGE",)
RETURN_NAMES = ("image",)
@@ -35,7 +35,6 @@ class Waifu2x:
self.output_type = kwargs.get('output_type', 'pil')
self.scale = kwargs.get('scale', 1)
self.tile_size = kwargs.get('tile_size', 256)
self.load()
def load(self, model_type: Optional[str] = 'art'):
@@ -47,22 +46,21 @@ class Waifu2x:
keep_alpha=self.keep_alpha,
amp=self.amp,
trust_repo=True,
).to('cuda')
)
self.method = self.waifu2x_method(self.scale, self.noise_level)
self.model.set_mode(method=self.method, noise_level=self.noise_level)
def set(
self,
amp: Optional[bool] = None,
batch_size: Optional[int] = None,
enabled: Optional[bool] = None,
keep_alpha: Optional[bool] = None,
model_type: Optional[str] = None,
noise_level: Optional[int] = None,
output_type: Optional[str] = None,
scale: Optional[int] = None,
tile_size: Optional[int] = None,
self,
amp: Optional[bool] = None,
batch_size: Optional[int] = None,
enabled: Optional[bool] = None,
keep_alpha: Optional[bool] = None,
model_type: Optional[str] = None,
noise_level: Optional[int] = None,
output_type: Optional[str] = None,
scale: Optional[int] = None,
tile_size: Optional[int] = None,
):
settings = locals()
settings.pop('self')
@@ -86,36 +84,53 @@ class Waifu2x:
return 'noise_scale4x'
def __call__(
self,
image,
scale: Optional[int] = 1,
noise_level: Optional[int] = 3,
model_type: Optional[str] = 'art',
self,
image,
scale: Optional[int] = 1,
noise_level: Optional[int] = 3,
model_type: Optional[str] = 'art',
**kwargs
):
print(f'Waifu2x Upscaling image with noise_level:{noise_level} scale:{scale}x model_type:{model_type}...')
print(f'Type of image: {type(image)}')
print('WAIFU2X CALL:', scale, noise_level, model_type)
if self.model_type != model_type:
self.load(model_type)
image = convert(image)
scale = scale if scale is not None else self.scale
noise_level = noise_level if noise_level is not None else self.noise_level
if isinstance(image, torch.Tensor):
if image.dim() == 3:
image = [image]
elif image.dim() == 4:
image = [img for img in image]
if self.scale != scale or self.noise_level != noise_level:
self.method = self.waifu2x_method(scale, noise_level)
self.model.set_mode(self.method, noise_level)
image = self.model.infer(image, method=self.method, noise_level=noise_level, output_type='pil')
image = convert(image)
return (image,)
tensors = []
for img in image:
img = img.cpu()
img_pil = tensor2pil(img)
# @classmethod
# def IS_CHANGED(self, video, **kwargs):
# return hash_path(video)
#
# @classmethod
# def VALIDATE_INPUTS(self, video, **kwargs):
# return validate_path(video, allow_none=True)
scale = scale if scale is not None else self.scale
noise_level = noise_level if noise_level is not None else self.noise_level
if self.scale != scale or self.noise_level != noise_level:
self.method = self.waifu2x_method(scale, noise_level)
self.model.set_mode(self.method, noise_level)
print('METHOD:', self.method)
img_pil = self.model.infer(img_pil, method=self.method, noise_level=noise_level, output_type='pil')
img_tensor = pil2tensor(img_pil)
tensors.append(img_tensor)
if len(tensors) > 1:
output = torch.cat(tensors, dim=0)
else:
output = tensors[0]
if output.dtype != torch.float32:
output = output.float()
if output.max() > 1.0:
output = output / 255.0
return (output,)
NODE_CLASS_MAPPINGS = {"Waifu2x": Waifu2x}
requirements.txt
+2
View File
@@ -0,0 +1,2 @@
colour-science
rembg
utils/convert.py
+10
View File
@@ -16,3 +16,13 @@ def convert(image):
image = Image.fromarray(np.clip(img_array, 0, 255).astype(np.uint8))
return image
# Tensor to PIL
def tensor2pil(image):
return Image.fromarray(np.clip(255. * image.cpu().numpy().squeeze(), 0, 255).astype(np.uint8))
# PIL to Tensor
def pil2tensor(image):
return torch.from_numpy(np.array(image).astype(np.float32) / 255.0).unsqueeze(0)
web/index.js
+24
View File
@@ -0,0 +1,24 @@
import { app } from '../../scripts/app.js';
import { loadCSS, loadScript, sleep } from './utils.js';
import { applyStyles } from './node-styles.js';
import { showCode } from './show-code.js';
import { storeAnyNodeOutputs } from './store-outputs.js';
app.registerExtension({
name: 'ShibikoAI',
async init() {
const CDN_BASE_URL = 'https://cdnjs.cloudflare.com/ajax/libs';
await loadScript(`${CDN_BASE_URL}/highlight.js/11.9.0/highlight.min.js`);
await loadScript(`${CDN_BASE_URL}/highlightjs-line-numbers.js/2.8.0/highlightjs-line-numbers.min.js`);
await loadCSS(`${CDN_BASE_URL}/highlight.js/11.9.0/styles/atom-one-dark.min.css`);
await loadCSS(`/extensions/anynode/css/show-code.css`);
hljs.configure({ ignoreUnescapedHTML: true });
},
async nodeCreated(node) {
await sleep(0); // Wait for node object to be updated
if (node.type.includes('AnyNode')) {
applyStyles(node);
node.type === 'AnyNodeShowCode' ? showCode(node) : storeAnyNodeOutputs(node);
}
},
});