63 lines
1.9 KiB
Markdown
63 lines
1.9 KiB
Markdown
|
|
# Documentation
|
|
- Class name: ImageAlphaComposite
|
|
- Category: Art Venture/Utils
|
|
- Output node: False
|
|
- Repo Ref: https://github.com/comfyanonymous/ComfyUI
|
|
|
|
ImageAlphaComposite节点用于根据两张图片的alpha通道值将它们混合在一起，生成一张合成图像。这个过程涉及将两张输入图像的视觉元素组合成一张图像，同时考虑透明度和图层效果。
|
|
|
|
# Input types
|
|
## Required
|
|
- image_i
|
|
- 参与合成的第二张图像。它作为alpha合成过程中的另一个基础层，与第一张图像叠加在一起。
|
|
- Comfy dtype: IMAGE
|
|
- Python dtype: torch.Tensor
|
|
|
|
# Output types
|
|
- image
|
|
- 经过alpha合成后得到的结果图像，基于输入图像的alpha值将它们混合在一起。
|
|
- Comfy dtype: IMAGE
|
|
- Python dtype: torch.Tensor
|
|
|
|
|
|
## Usage tips
|
|
- Infra type: `GPU`
|
|
- Common nodes: unknown
|
|
|
|
|
|
## Source code
|
|
```python
|
|
class UtilImageAlphaComposite:
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return {
|
|
"required": {
|
|
"image_1": ("IMAGE",),
|
|
"image_2": ("IMAGE",),
|
|
}
|
|
}
|
|
|
|
RETURN_TYPES = ("IMAGE",)
|
|
CATEGORY = "Art Venture/Utils"
|
|
FUNCTION = "image_alpha_composite"
|
|
|
|
def image_alpha_composite(self, image_1: torch.Tensor, image_2: torch.Tensor):
|
|
if image_1.shape[0] != image_2.shape[0]:
|
|
raise Exception("Images must have the same amount")
|
|
|
|
if image_1.shape[1] != image_2.shape[1] or image_1.shape[2] != image_2.shape[2]:
|
|
raise Exception("Images must have the same size")
|
|
|
|
composited_images = []
|
|
for i, im1 in enumerate(image_1):
|
|
composited = Image.alpha_composite(
|
|
tensor2pil(im1).convert("RGBA"),
|
|
tensor2pil(image_2[i]).convert("RGBA"),
|
|
)
|
|
composited_images.append(pil2tensor(composited))
|
|
|
|
return (torch.cat(composited_images, dim=0),)
|
|
|
|
```
|