imgutils.edge.lineart

Overview:

Get edge with lineart model.

Having the best effect, closest to the drawing lines, but consuming a large amount of memory and computing power at runtime.

get_edge_by_lineart

imgutils.edge.lineart.get_edge_by_lineart(image: str | PathLike | bytes | bytearray | BinaryIO | Image, coarse: bool = False, detect_resolution: int = 512)

Overview:

Get edge mask with lineart model.

Parameters:

• image – Original image (assuming its size is HxW).

• coarse – Use coarse model. In the coarse model, the lines will be deeper and richer, but the probability of extra lines or content appearing will increase.

• detect_resolution – Resolution when passing the image into neural network. Default is 512.

Returns:

A mask with format float32[H, W].

edge_image_with_lineart

imgutils.edge.lineart.edge_image_with_lineart(image: str | PathLike | bytes | bytearray | BinaryIO | Image, coarse: bool = False, detect_resolution: int = 512, backcolor: str = 'white', forecolor: str | None = None)

Overview:

Get an image with the extracted edge from image.

Parameters:

• image – Original image (assuming its size is HxW).

• coarse – Use coarse model. In the coarse model, the lines will be deeper and richer, but the probability of extra lines or content appearing will increase.

• detect_resolution – Resolution when passing the image into neural network. Default is 512.

• backcolor – Background color the target image. Default is white. When transparent is given, the background will be transparent.

• forecolor – Fore color of the target image. Default is None which means use the color from the given image.

Returns:

An image with the extracted edge from image.

Examples::

When coarse is used: