This past weekend was the first time in five weekends that I was in town. I was going to get all sorts of work done around the house, etc and catch up on things. Photography was still going to be relegated to the wish list — no time for that. Well, Thursday night I started to feel a bit under the weather and by Friday morning I was out-and-out ill. I ended up in bed throughout this weekend and one of the things I did (when not in a complete fog) was watch a few videos on kelbytraining.com. I got to do *something* related to photography at least.
Being the geek that I am, I watched a few videos on the Lab color space done by a guy named Dan Margulis. Lab is a color space (like RGB for example) which uses three channels: ‘L’ for luminosity, ‘a’ for green/magenta, and ‘b’ for blue/yellow. I won’t even try to explain when and why one might want to use the Lab color space but I will attempt a poor-man’s explanation of one use I’ve already found for Lab with the help of Mr. Margulis.
The portrait above-left was taken on a bright, sunny day in Vancouver’s Stanley Park. Note the blown-out highlights on the forehead and nose. No amount of curves, saturation, or other adjustments would bring the proper color back to those spots. The only real option is to paint color in if you really want the color back. The way you’d generally do this is to sample a nearby color in the skin (option-click when you have the paintbrush tool open in Photoshop) and paint over the area. The main problem with this is that your “paint” covers every pixel — better make sure you don’t paint over any of the areas you want to keep (those areas with non-blown-out colors and textures). One way to prevent this is to paint on another layer and use the ‘color’ blend mode which will add color but keep the existing texture. The problem with this? In RGB, adding color to something already blown-out will simply remain blown-out.
If you convert your layer to Lab (I’m not going to attempt to explain the exact mechanics because I’ll surely leave out something important), you can paint while using color blend mode and the color “sticks”. This is essentially because Lab has a much larger color gamut than RGB. One way of saying it is that in Lab space there exists a color which is blown-out (from a luminosity standpoint) but still has a color value. Your first thought might be, “Don’t you lose that color when you convert back to RGB eventually?”. Nope. And that’s just the way it works — Photoshop doesn’t know that the Lab colors you painted in were blown-out highlights. It just sees a color that it has to make a best guess about converting back to RGB.
It’s very subtle, but in the above-right image you can see that I added a touch of color to the blown-out spots on the forehead and nose, and to the right ear and forehead above the right eye. There are still highlights, but they are no longer brilliant white.
Here’s another example. In this sunset silhouette taken in Corpus Christi, TX there’s a huge blown-out area in the sky. In RGB you could paint some color in very carefully — being sure to avoid the buildings, etc. In Lab, I simply painted in color using the ‘color’ blend mode and the image is much-improved IMO. There’s now a touch of color in the whole sky and in the water and the edit took about one minute. [You'll note that the overall color cast is slightly different on the right and I simply don't remember if I touched some other setting -- Just trust that the color in the previously blown-out areas is due to painting in the Lab color space.]
I have no doubt that there are 20 other ways to tackle the problem of blown-out highlights in post, but I wanted to share this one that I learned. If you’re geeky enough to find this interesting, I hope I’ve whet your appetite enough to go figure it out with the help of a book or video. If you’re not geeky enough then I won’t be able to explain it well enough to help you anyway.