如何使用Python实现类似Photoshop“斜角/浮雕”的深度效果？

Question

我想将图像放置在背景图像之上，并使用 Python 在图像上应用“斜角/浮雕”效果。我尝试使用 PIL 库，但也欢迎对其他库提出建议。

它应该是这样的：

我有以下代码：

from PIL import Image
from PIL import ImageFilter

img = Image.open('./image.jpeg', 'r')

# this doesn't do what I want...
img = img .filter(ImageFilter.EMBOSS)

background = Image.open('./bg.png', 'r')

background.paste(img)

我使用 Affinity Photo 作为示例图像。在 Photoshop 中应该是差不多的。以下是我使用的设置：

Answer 1

我仍然不知道如何在 python 中执行此操作，但我找到了一种方法，结合使用批处理作业和 Affinity Photo 中的宏功能。

1. 录制一个宏，将所需的效果应用于图像
2. 将所有需要效果的图片放在一个文件夹中
3. 启动批处理作业并将宏应用于所有图像

此处描述了如何启动批处理作业：http://www.millermattson.com/blog/batch-processing-with-affinity-photo/

Answer 2

我在谷歌搜索试图弄清楚如何使用 PIL 对文本进行斜角和浮雕时发现了这个问题。这花了几个小时的时间，但我对解决方案非常满意并想分享：

def draw_embossed_text(image, text, font, position, fill, highlight, shadow):
    """
    Create an embossed text effect on an image

    :param image: PIL Image object to write on
    :param text: The text to write
    :param font: The ImageFont for the text
    :param position: Tuple (x, y) of the position to write the text
    :param fill: The color of the text as an RGB tuple
    :param highlight: Highlight color as an RGB tuple
    :param shadow: Shadow color as an RGB tuple
    :return: Modified PIL Image object with embossed text
    """
    # Convert image to RGBA
    image = image.convert('RGBA')

    # Create text mask
    text_mask = Image.new('L', image.size, 0)
    draw = ImageDraw.Draw(text_mask)
    draw.text(position, text, font=font, fill=255)

    # Create a layer for the fill color text
    fill_text_layer = Image.new('RGBA', image.size, (0, 0, 0, 0))
    fill_draw = ImageDraw.Draw(fill_text_layer)
    fill_draw.text(position, text, font=font, fill=fill + (255,))

    # Apply emboss filter to the text mask
    embossed_mask = text_mask.filter(ImageFilter.EMBOSS)
    embossed_mask = ImageOps.autocontrast(embossed_mask).filter(ImageFilter.GaussianBlur(3))

    # Create a layer for the embossed effect using the mask
    embossed_layer = Image.new('RGBA', image.size, fill + (0,))
    embossed_layer.putalpha(embossed_mask)

    # Convert embossed layer to NumPy array for pixel manipulation
    embossed_array = np.array(embossed_layer)
    alpha_channel = embossed_array[:, :, 3]

    # Determine highlight and shadow areas
    highlight_area = alpha_channel > 128
    shadow_area = alpha_channel < 128

    # Apply highlight and shadow
    embossed_array[highlight_area, :3] = highlight
    embossed_array[shadow_area, :3] = shadow

    # Adjust intensity
    intensity = np.minimum(np.minimum(np.abs(alpha_channel - 128), np.abs(128 - alpha_channel)) * 2, 255)
    embossed_array[:,:,3] = np.where(highlight_area | shadow_area, intensity, 0)

    # Convert back to PIL Image
    embossed_layer = Image.fromarray(embossed_array)
    
    # Composite the embossed text onto the original image
    image_with_text = Image.alpha_composite(image, fill_text_layer)
    final_image = Image.alpha_composite(image_with_text, embossed_layer)

    return final_image