More fun with the PIL

That's the Python Imaging Library. I demonstrated a couple of edge detection techniques (gradient and laplacian) using the PIL to simply read in the image files and make the pixel values conveniently available for twiddling. There's lots of neat stuff in the library other than just image reading/writing utilities. The blend method for instance.
I created a gray-scale image of the magnitude of the gradient (the edges in the image), and then blended it with the original color image. The python to do that takes just two lines:

cblended = Image.blend(outimg.convert("RGB"), img, 0.4)
cblended.save("kuai_laying_color_blend.jpg")

Image.blend() returns an image that is a linear interpolation of the first two arguments (images). The third argument controls if the emphasis is placed on the first image (number closer to zero) or the second image (number closer to one).

Makes it look kind of like a water color, no?