Digital Solarization

If you spent much time in a black-and-white darkroom, you probably at least experimented with the solarization process. Here's how it worked: 

1. You printed a picture normally.
2. You developed it, leaving it face up in the developer tray.
3. You re-exposed the picture to light while it was in the developer tray. (This usually meant just turning on the lights for several seconds.)
4. You let the picture continue to develop for another minute or so.
5. You finished processing the picture in the normal way.

If the darkroom gods were kind to you, you got a nice solarized print out of this. If not, you tried again, with maybe more or less re-exposure, a lighter print to start with, or something. The process was really not very easy to control.

A solarized print has two distinguishing characteristics. The first is that the dark values look pretty much normal while the light values are reversed, as in a negative. This has come to be known as the Sabattier effect. The second is that light-valued lines appear in the solarization where the original image has edges between areas of light and dark values. These are called Mackie lines.

I took a notion one day to write yet another computer program: This one to produce a solarized-looking image from a normal image. Emulating the Sabattier effect is fairly straightforward. You leave the values below middle gray alone, and you reverse the values above middle gray.

Mackie lines can be thought of as an "edges" image that comes out of the Finding Edges process. (See the Finding Edges article.) Running the Finding Edges process on the original image and then overlaying the resulting edges on the Sabattier-effect image gives convincing Mackie lines.

What does it look like? Here's an original image:

And here is the image run through the Digital Solarization process:

My program converts the original image to grayscale and produces a traditional-looking grayscale solarization. You can do color solarization by running the process on each color channel separately, but the results don't seem pleasing to me.