117 lines
4.3 KiB
Python
117 lines
4.3 KiB
Python
"""
|
|
XB-ToolBox 批量缩放图像节点
|
|
=============================
|
|
复刻官方 ImageScale，支持动态输入/输出（最多9张）。
|
|
N进N出，共享缩放参数，每张独立处理，互不干扰。
|
|
"""
|
|
|
|
import math
|
|
import comfy.utils
|
|
import nodes
|
|
from .nodes_reference_any import FlexibleOptionalInputType
|
|
|
|
MAX_IMAGES = 9
|
|
|
|
|
|
class XB_ImageScale:
|
|
"""批量缩放图像 — 动态输入/输出，N进N出。"""
|
|
|
|
upscale_methods = ["nearest-exact", "bilinear", "area", "bicubic", "lanczos"]
|
|
crop_methods = ["disabled", "center"]
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
return {
|
|
"required": {
|
|
"upscale_method": (cls.upscale_methods,),
|
|
"width": ("INT", {"default": 512, "min": 0, "max": nodes.MAX_RESOLUTION, "step": 1}),
|
|
"height": ("INT", {"default": 512, "min": 0, "max": nodes.MAX_RESOLUTION, "step": 1}),
|
|
"crop": (cls.crop_methods,),
|
|
"image1": ("IMAGE",),
|
|
},
|
|
"optional": FlexibleOptionalInputType("IMAGE"),
|
|
}
|
|
|
|
RETURN_TYPES = ("IMAGE",) * MAX_IMAGES
|
|
RETURN_NAMES = tuple(f"图像{i+1}" for i in range(MAX_IMAGES))
|
|
FUNCTION = "upscale"
|
|
CATEGORY = "XB_ToolBox/Utils"
|
|
|
|
def upscale(self, upscale_method, width, height, crop, image1, **kwargs):
|
|
# ── 收集所有图片 ──
|
|
images = [image1]
|
|
for key, value in kwargs.items():
|
|
if value is not None and hasattr(value, 'movedim'):
|
|
images.append(value)
|
|
|
|
# ── 逐张缩放 ──
|
|
results = []
|
|
for img in images:
|
|
if width == 0 and height == 0:
|
|
results.append(img)
|
|
else:
|
|
samples = img.movedim(-1, 1)
|
|
w, h = width, height
|
|
if w == 0:
|
|
w = max(1, round(samples.shape[3] * h / samples.shape[2]))
|
|
elif h == 0:
|
|
h = max(1, round(samples.shape[2] * w / samples.shape[3]))
|
|
s = comfy.utils.common_upscale(samples, w, h, upscale_method, crop)
|
|
results.append(s.movedim(1, -1))
|
|
|
|
# ── 填充到 MAX_IMAGES 个输出 ──
|
|
fallback = results[0] if results else image1
|
|
while len(results) < MAX_IMAGES:
|
|
results.append(fallback)
|
|
|
|
return tuple(results)
|
|
|
|
|
|
class XB_ImageScalePixels:
|
|
"""批量缩放图像（像素）— 按总像素数缩放，动态输入/输出，N进N出。"""
|
|
|
|
upscale_methods = ["nearest-exact", "bilinear", "area", "bicubic", "lanczos"]
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
return {
|
|
"required": {
|
|
"upscale_method": (cls.upscale_methods,),
|
|
"megapixels": ("FLOAT", {"default": 1.0, "min": 0.01, "max": 16.0, "step": 0.01}),
|
|
"resolution_step": ("INT", {"default": 1, "min": 1, "max": 1024, "step": 1,
|
|
"tooltip": "宽高对齐到该步长的整数倍"}),
|
|
"image1": ("IMAGE",),
|
|
},
|
|
"optional": FlexibleOptionalInputType("IMAGE"),
|
|
}
|
|
|
|
RETURN_TYPES = ("IMAGE",) * MAX_IMAGES
|
|
RETURN_NAMES = tuple(f"图像{i+1}" for i in range(MAX_IMAGES))
|
|
FUNCTION = "upscale"
|
|
CATEGORY = "XB_ToolBox/Utils"
|
|
|
|
def upscale(self, upscale_method, megapixels, resolution_step, image1, **kwargs):
|
|
images = [image1]
|
|
for key, value in kwargs.items():
|
|
if value is not None and hasattr(value, 'movedim'):
|
|
images.append(value)
|
|
|
|
total = int(megapixels * 1024 * 1024)
|
|
results = []
|
|
for img in images:
|
|
samples = img.movedim(-1, 1)
|
|
scale_by = math.sqrt(total / (samples.shape[3] * samples.shape[2]))
|
|
w = max(1, round(samples.shape[3] * scale_by))
|
|
h = max(1, round(samples.shape[2] * scale_by))
|
|
if resolution_step > 1:
|
|
w = max(resolution_step, int(round(w / resolution_step)) * resolution_step)
|
|
h = max(resolution_step, int(round(h / resolution_step)) * resolution_step)
|
|
s = comfy.utils.common_upscale(samples, w, h, upscale_method, "disabled")
|
|
results.append(s.movedim(1, -1))
|
|
|
|
fallback = results[0] if results else image1
|
|
while len(results) < MAX_IMAGES:
|
|
results.append(fallback)
|
|
|
|
return tuple(results)
|