64 lines
1.9 KiB
Markdown
64 lines
1.9 KiB
Markdown
|
|
# Documentation
|
|
- Class name: ImageExtractChannel
|
|
- Category: Art Venture/Utils
|
|
- Output node: False
|
|
|
|
ImageExtractChannel节点专门用于从给定的一组图像中提取特定的颜色通道（红色、绿色、蓝色或阿尔法）。它允许对单个颜色通道进行操作和分析，这对于各种图像处理任务至关重要，如创建蒙版或隔离颜色组件。
|
|
|
|
# Input types
|
|
## Required
|
|
- images
|
|
- 将从中提取特定通道的图像。这个输入对于确定要处理的源图像至关重要。
|
|
- Comfy dtype: IMAGE
|
|
- Python dtype: torch.Tensor
|
|
- channel
|
|
- 指定要从图像中提取的颜色通道（红色、绿色、蓝色或阿尔法）。这个选择通过隔离所需的通道直接影响输出。
|
|
- Comfy dtype: COMBO[STRING]
|
|
- Python dtype: str
|
|
|
|
# Output types
|
|
- channel_data
|
|
- 从输入图像中提取的通道数据，以蒙版形式提供。这个输出对于进一步的图像处理或分析任务非常有用。
|
|
- Comfy dtype: MASK
|
|
- Python dtype: torch.Tensor
|
|
|
|
|
|
## Usage tips
|
|
- Infra type: `GPU`
|
|
- Common nodes: unknown
|
|
|
|
|
|
## Source code
|
|
```python
|
|
class UtilImageExtractChannel:
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return {
|
|
"required": {
|
|
"images": ("IMAGE",),
|
|
"channel": (["R", "G", "B", "A"],),
|
|
}
|
|
}
|
|
|
|
RETURN_TYPES = ("MASK",)
|
|
RETURN_NAMES = ("channel_data",)
|
|
CATEGORY = "Art Venture/Utils"
|
|
FUNCTION = "image_extract_alpha"
|
|
|
|
def image_extract_alpha(self, images: torch.Tensor, channel):
|
|
# images in shape (N, H, W, C)
|
|
|
|
if len(images.shape) < 4:
|
|
images = images.unsqueeze(3).repeat(1, 1, 1, 3)
|
|
|
|
if channel == "A" and images.shape[3] < 4:
|
|
raise Exception("Image does not have an alpha channel")
|
|
|
|
channel_index = ["R", "G", "B", "A"].index(channel)
|
|
mask = images[:, :, :, channel_index].cpu().clone()
|
|
|
|
return (mask,)
|
|
|
|
```
|