Performing Photoshop's "Luminosity" filter programmatically

Question

I have two JPEG's and would like to overlay one on the other with the same results as the "Luminosity" mode available in Photoshop (and Fireworks). You can read more about Luminosity mode here: http://www.adobetutorialz.com/articles/662/1/Photoshop%92s-Luminosity-Mode

How can I do this? Programming language doesn't matter much, but I am most fluent with Python and PHP (in that order). Python Imaging Library seems like a perfect fit, but luminosity is not a built-in function and I do not know the proper procedure. See http://www.pythonware.com/library/pil/handbook/imagechops.htm

Answer 1

The Gimp would be another option - it has a scripting interface ans a python api - here is an article on luminosity and the Gimp. Not sure if it is the same effect you are going for though.

Answer 2

First you need to understand what Photoshop does.

It preserves under layer perceptual color information and replaces it's luminosity with the top layer's perceptual luminosity information. To do that, you need to convert the images to the right color space.

Here is the shoping list of things you will need to do if you decide to implement everything by yourself:

• Load both the source and target JPEGs
• Convert the pixels from RGB color space to L*a*b color space (or any other color space with luminosirty information)
• Preserve target color channels and replace its luminosity channel by source's luminosity
• Convert back to RGB space
• Save the JPEG

If you think Lab is too complicated, you can also use HSL color space, it's much simpler but it will give inferior results.

Answer 3

In pseudo-code:

foreach rgb_pixel1, rgb_pixel2 in image1, image2 {
    hsl1 = RgbToHsl(rgb_pixel1);
    hsl2 = RgbToHsl(rgb_pixel2);
    hsl3 = hsl(hsl1.h, hsl1.s, hsl2.l);
    output_rgb = HslToRgb(hsl3);
}

Conversion from rgb to hsl and back is here.

Answer 4

You could have a look at the OpenCV image processing library. It has Python bindings and handles a lot of these lower level image manipulation tasks for you, or at least makes them easier.

Answer 5

I don't know about this specific filter but I can tell you how to follow Coincoin steps in PIL. I didn't actually run the code, but you can use it as a reference:

Load both the source and target JPEGs

from PIL import Image
img1 = Image.open('image1.jpg')
img2 = Image.open('image2.jpg')

Convert the pixels from RGB color space to L*a*b color space (or any other color space with luminosirty information)

# Color matrix for Lab
colorMatrix = (
    x1, y1, z1, 0,
    x2, y2, z2, 0,
    x3, y3, z3, 0
)
img1 = img1.convert("RGB", colorMatrix)
img2 = img2.convert("RGB", colorMatrix)

Preserve target color channels and replace its luminosity channel by source's luminosity

l1, a1, b1 = img1.split()
l2, a2, b2 = img2.split()
img1.putdata(zip(l1.getdata(), a2.getdata(), b2.getdata()))

Convert back to RGB space

# Color matrix for RGB
RGBcolorMatrix = (
    x1, y1, z1, 0,
    x2, y2, z2, 0,
    x3, y3, z3, 0
)
img1 = img1.convert("RGB", RGBcolorMatrix)

Save the JPEG

img1.save('new_image.jpg')