What's the formula used in PIL.ImageEnhance enhance feature for color, brightness and contrast?

Question

How are the images transformed when using the PIL.ImageEnhance enhance feature for brightness, color and contrast respectively? As in, what is the mathematical formula for transforming the pixel values for each of the transformations?

Answer 1

From the source code on ImageEnhance.py , we see that there is no direct "mathematical formula" for any of the transformations. For each modality (brightness, color, etc.), a modified image is generated and then this modified image is blended with the original image by the given factor from 0.0 ... 1.0 , cf. Image.blend .

• Brightness: Original image is blended with a black image of the same size.
• Color: Original image is blended with its grayscale version.
• Contrast: Original image is blended with a gray image of the same size. Here's the only point, where some calculation takes place. The gray value is determined by the mean of the original image' grayscale version.

Here's some comparison code, rebuilding the named functions from ImageEnhancer on-the-fly for some RGB image:

from matplotlib import pyplot as plt
from PIL import Image, ImageEnhance, ImageStat

# Read image, set up factor
image = Image.open('path/to/your/image.png')
factor = 0.25

# ImageEnhance
br_enhancer = ImageEnhance.Brightness(image)
cl_enhancer = ImageEnhance.Color(image)
cn_enhancer = ImageEnhance.Contrast(image)

# Rebuild ImageEnhance.Brightness on-the-fly
br_image_pre = Image.new(image.mode, image.size, 0)
br_image = Image.blend(br_image_pre, image, factor)

# Rebuild ImageEnhance.Color on-the-fly
cl_image_pre = image.convert('L').convert('RGB')
cl_image = Image.blend(cl_image_pre, image, factor)

# Rebuild ImageEnhance.Contrast on-the-fly
mean = int(ImageStat.Stat(image.convert('L')).mean[0] + 0.5)
cn_image_pre = Image.new('L', image.size, mean).convert(image.mode)
cn_image = Image.blend(cn_image_pre, image, factor)

# Visualization
plt.figure(1, figsize=(14, 9))
plt.subplot(3, 4, 1), plt.imshow(image), plt.title('Original image')
plt.subplot(3, 4, 2), plt.imshow(br_enhancer.enhance(factor)), plt.title('ImageEnhance.Brightness(0.25)')
plt.subplot(3, 4, 3), plt.imshow(cl_enhancer.enhance(factor)), plt.title('ImageEnhance.Color(0.25)')
plt.subplot(3, 4, 4), plt.imshow(cn_enhancer.enhance(factor)), plt.title('ImageEnhance.Contrast(0.25)')
plt.subplot(3, 4, 5), plt.imshow(image), plt.title('Original image (0.25)')
plt.subplot(3, 4, 6), plt.imshow(br_image_pre), plt.title('+ brightness modified image (0.75)')
plt.subplot(3, 4, 7), plt.imshow(cl_image_pre), plt.title('+ color modified image (0.75)')
plt.subplot(3, 4, 8), plt.imshow(cn_image_pre), plt.title('+ contrast modified image (0.75)')
plt.subplot(3, 4, 10), plt.imshow(br_image), plt.title('= rebuilt ImageEnhance.Brightness(0.25)')
plt.subplot(3, 4, 11), plt.imshow(cl_image), plt.title('= rebuilt ImageEnhance.Color(0.25)')
plt.subplot(3, 4, 12), plt.imshow(cn_image), plt.title('= rebuilt ImageEnhance.Contrast(0.25)')
plt.tight_layout()
plt.show()

And, that's the output for my standard test image:

Hope that helps understanding!