domg73
e490a44ef2
- New 'flip' input (None/Horizontal/Vertical, default None).
- Selecting a flip mirrors the image in the browser (canvas), saves it
to input/ as {name}-flipH.png / {name}-flipV.png, and switches the
node's image input to the flipped file.
- The flipped image is used everywhere (preview, mask editor, output);
the crop rectangle is framed on the flipped image (WYSIWYG).
- 'None' reverts to the original image.
- No canvas rendering tricks: the flip is a real file in input/, so it
is consistent across all contexts.
Files changed:
- load_image_crop.py: FLIP_OPTIONS + flip input in INPUT_TYPES
- web/js/loadImageCrop.js: applyFlip() + flip widget binding
- pyproject.toml: version 1.0.7
- CHANGELOG.md: v1.0.7 entry
- README.md: Flip section updated
175 lines
7.0 KiB
Python
175 lines
7.0 KiB
Python
import os
|
|
import hashlib
|
|
import numpy as np
|
|
import torch
|
|
from PIL import Image, ImageOps, ImageSequence
|
|
|
|
import folder_paths
|
|
import comfy
|
|
import node_helpers
|
|
|
|
try: # current import location
|
|
from comfy_api.latest import InputImpl
|
|
except Exception:
|
|
try: # older import location
|
|
from comfy.api.latest import InputImpl
|
|
except Exception:
|
|
InputImpl = None
|
|
|
|
ASPECT_RATIOS = [
|
|
"Original",
|
|
"Free (Custom)",
|
|
"1:1 (Square)",
|
|
"2:3 (Portrait Photo)",
|
|
"3:2 (Photo)",
|
|
"3:4 (Portrait Standard)",
|
|
"4:3 (Standard)",
|
|
"9:16 (Portrait Widescreen)",
|
|
"16:9 (Widescreen)",
|
|
"21:9 (Ultrawide)",
|
|
]
|
|
|
|
FLIP_OPTIONS = [
|
|
"None",
|
|
"Horizontal",
|
|
"Vertical",
|
|
]
|
|
|
|
|
|
class LoadImageCrop:
|
|
"""Like the official Load Image node, plus:
|
|
- a visual, mouse-draggable crop area (aspect-ratio locked) in the frontend preview
|
|
- a paste-from-clipboard entry in the node context menu
|
|
|
|
With aspect_ratio == "Original" (default) the image is returned as-is,
|
|
uncropped.
|
|
|
|
Otherwise the crop rectangle (crop_x, crop_y, crop_w, crop_h) arrives as
|
|
normalized [0..1] fractions of the source image size, computed by the
|
|
frontend.
|
|
"""
|
|
|
|
CATEGORY = "image"
|
|
SEARCH_ALIASES = ["load image", "crop", "crop image", "paste", "clipboard"]
|
|
|
|
RETURN_TYPES = ("IMAGE", "MASK")
|
|
RETURN_NAMES = ("IMAGE", "MASK")
|
|
FUNCTION = "load_image"
|
|
OUTPUT_NODE = False
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
input_dir = folder_paths.get_input_directory()
|
|
files = [f for f in os.listdir(input_dir) if os.path.isfile(os.path.join(input_dir, f))]
|
|
files = folder_paths.filter_files_content_types(files, ["image"])
|
|
return {
|
|
"required": {
|
|
"image": (sorted(files), {"image_upload": True}),
|
|
"aspect_ratio": (ASPECT_RATIOS,),
|
|
"flip": (FLIP_OPTIONS, {"default": "None"}),
|
|
"crop_x": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.0001}),
|
|
"crop_y": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.0001}),
|
|
"crop_w": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.0001}),
|
|
"crop_h": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.0001}),
|
|
}
|
|
}
|
|
|
|
def load_image(self, image, aspect_ratio, flip, crop_x, crop_y, crop_w, crop_h):
|
|
output_image, output_mask = self._load(image)
|
|
if aspect_ratio == "Original":
|
|
# the image is returned as-is, regardless of the crop values
|
|
return (output_image, output_mask)
|
|
return self._crop(output_image, output_mask, crop_x, crop_y, crop_w, crop_h)
|
|
|
|
def _load(self, image):
|
|
"""Load the image (and its alpha mask) as tensors, without any crop."""
|
|
image_path = folder_paths.get_annotated_filepath(image)
|
|
|
|
dtype = comfy.model_management.intermediate_dtype()
|
|
device = comfy.model_management.intermediate_device()
|
|
|
|
if InputImpl is not None:
|
|
components = InputImpl.VideoFromFile(image_path).get_components()
|
|
if components.images.shape[0] > 0:
|
|
output_image = components.images.to(device=device, dtype=dtype)
|
|
if components.alpha is not None:
|
|
output_mask = (1.0 - components.alpha[..., -1]).to(device=device, dtype=dtype)
|
|
else:
|
|
output_mask = torch.zeros((output_image.shape[0], 64, 64), dtype=dtype, device=device)
|
|
return (output_image, output_mask)
|
|
|
|
# Fallback (also used when InputImpl is unavailable): plain image loading,
|
|
# the same path the official node uses for animated webp.
|
|
img = node_helpers.pillow(Image.open, image_path)
|
|
output_images, output_masks = [], []
|
|
w, h = None, None
|
|
for i in ImageSequence.Iterator(img):
|
|
i = node_helpers.pillow(ImageOps.exif_transpose, i)
|
|
rgb = i.convert("RGB")
|
|
if len(output_images) == 0:
|
|
w, h = rgb.size[0], rgb.size[1]
|
|
if rgb.size[0] != w or rgb.size[1] != h:
|
|
continue
|
|
arr = np.array(rgb).astype(np.float32) / 255.0
|
|
output_images.append(torch.from_numpy(arr)[None,].to(dtype=dtype))
|
|
if "A" in i.getbands():
|
|
mask = np.array(i.getchannel("A")).astype(np.float32) / 255.0
|
|
mask = 1.0 - torch.from_numpy(mask)
|
|
else:
|
|
mask = torch.zeros((64, 64), dtype=torch.float32)
|
|
output_masks.append(mask.unsqueeze(0).to(dtype=dtype))
|
|
if len(output_images) == 0:
|
|
raise RuntimeError("no usable frames in image: {}".format(image))
|
|
output_image = torch.cat(output_images, dim=0).to(device=device, dtype=dtype)
|
|
output_mask = torch.cat(output_masks, dim=0).to(device=device, dtype=dtype)
|
|
return (output_image, output_mask)
|
|
|
|
def _crop(self, output_image, output_mask, crop_x, crop_y, crop_w, crop_h):
|
|
# full-frame crop ("Original" mode): return as-is
|
|
if (crop_x, crop_y, crop_w, crop_h) == (0, 0, 1, 1):
|
|
return (output_image, output_mask)
|
|
# guard against degenerate values (API/workflow input): an empty or
|
|
# out-of-range crop is returned uncropped
|
|
if not (crop_w > 0.001 and crop_h > 0.001 and crop_x < 0.999 and crop_y < 0.999):
|
|
return (output_image, output_mask)
|
|
|
|
H, W = output_image.shape[1], output_image.shape[2]
|
|
mh, mw = output_mask.shape[1], output_mask.shape[2]
|
|
|
|
def clampi(v, lo, hi):
|
|
return max(lo, min(hi, int(round(v))))
|
|
|
|
# crop the image
|
|
x0 = clampi(crop_x * W, 0, W - 1)
|
|
y0 = clampi(crop_y * H, 0, H - 1)
|
|
x1 = clampi((crop_x + crop_w) * W, x0 + 1, W)
|
|
y1 = clampi((crop_y + crop_h) * H, y0 + 1, H)
|
|
output_image = output_image[:, y0:y1, x0:x1, :]
|
|
|
|
# crop the mask proportionally (its size may differ from the image size)
|
|
mx0 = clampi(crop_x * mw, 0, mw - 1)
|
|
my0 = clampi(crop_y * mh, 0, mh - 1)
|
|
mx1 = clampi((crop_x + crop_w) * mw, mx0 + 1, mw)
|
|
my1 = clampi((crop_y + crop_h) * mh, my0 + 1, mh)
|
|
output_mask = output_mask[:, my0:my1, mx0:mx1]
|
|
|
|
return (output_image, output_mask)
|
|
|
|
@classmethod
|
|
def IS_CHANGED(s, image, aspect_ratio, flip, crop_x, crop_y, crop_w, crop_h):
|
|
# file identity via stat (size + mtime) instead of hashing the whole file
|
|
image_path = folder_paths.get_annotated_filepath(image)
|
|
st = os.stat(image_path)
|
|
m = hashlib.sha256()
|
|
m.update(str(st.st_size).encode())
|
|
m.update(str(st.st_mtime_ns).encode())
|
|
# full-precision crop values (repr) so any change forces a re-run
|
|
m.update(f"|{aspect_ratio}|{flip}|{crop_x!r},{crop_y!r},{crop_w!r},{crop_h!r}".encode())
|
|
return m.digest().hex()
|
|
|
|
@classmethod
|
|
def VALIDATE_INPUTS(s, image, aspect_ratio=None, flip=None, crop_x=None, crop_y=None, crop_w=None, crop_h=None):
|
|
if not folder_paths.exists_annotated_filepath(image):
|
|
return "Invalid image file: {}".format(image)
|
|
return True
|