The Photoshop problem is that the program is so rich in features, there are many ways of doing the same job and the most evident are not always the best. What follows is an overview of some of the techniques I’ve discovered on the web for converting colour photos to quality grayscale images.
The goal: to find a relatively simple technique that produces quality results without having to spend too much time (if any) with levels or curves. I haven’t tried any of the following with either low- or high-key images; I will try to get to those some time. Note: All of the following examples are straight conversions: no levels or curves have been applied to the final image.
And so we begin:
Simply changing the mode from RGB to greyscale (Image>Mode>grayscale) produces a flattish image that would require some work with levels or curves to improve. It doesn’t show in this photo, but I’ve found it particularly troublesome when dealing with fleshtones, less-than-good lighting, etc.
Going to LAB
Converting the image from RGB to LAB colour and then selecting the Lightness channel gives you access to the basic grayscale image. Conversions to grayscale from the Lightness channel (make sure it’s the only one that’s selected when you choose Image>Mode>Grayscale) often create a better starting point than a straight RGB to grayscale conversion. Not always though: this needs some work with curves.
This conversion has been done by a PhotoShop action (Custom RGB to Grayscale) that takes a snapshot of the image, opens the channel mixer, chooses monochrome, sets the values to +24 for red, +68 for green and +8 for blue. The channel mixer is left open, so you can make adjustments. The Channel Mixer allows you to decide how much of each of the channels you’ll use to create the final grayscale image: this action seems to do a good job of providing a starting point.
The result isn’t bad: the foreground is better than the mode conversion version, and the waters of Crater Lake are a little more interesting.
There’s a nice screencast dealing with the channel mixer here.
This is a much more complex way of converting images, but building an action that automates the various steps makes for quick conversion, while also leaving room for some tweaking, again without having to resort to levels or curves.
I’ve come across this technique from a couple of photographers, such as Greg Gorman and Martin Fuchs. The number of steps looks overwhelming, but don’t be put off by that. Download the action that both Gorman and Fuchs provide.
The key to the process is the fill layer. Once the action has run, you can access the fill layer (through the layers palette) and adjust the target tone. I like the way this action gave me a good foreground and better range in the water.
This is a variation on Channel Mixing, where you do the work. It involves duplicating the image (Image>Duplicate), deleting the image from the Duplicate (Select>All; Delete), and then copying selected channels from the original image onto layers in the duplicated, now empty image. I chose the red channel, which had the most detail as one layer, and the green channel, which had more detail and better tones for the foreground.
Once the layers were in the duplicated image, I set blend mode to multiply and then changed the opacity of the green layer to 36%. Using a blend mode of overlay produced great midtones but even at low opacity, there was too much contrast in the foreground. Choose channels carefully and play with the combination of blend modes and opacity. (See link below for more details on this technique.)
This photo is more dramatic, but the detail in the far shore is good, the foreground is fairly good and the sky a little more dramatic. The image is a little dark overall, leading to the next method.
Using a mask
Adding one step to the blended layer method above — adding a layer mask — gives me much more control and better results. After blending my red and green layers, I created a copy of the red layer, added a layer mask to that layer and then painted black over the bottom two-thirds of the image (on the layer mask) to lighten that part of the image, while maintaining the strong contrast of the cliffs and the sky. Looking at it now, it still needs a little work in the bottom third of the image, but you get the idea.
Creating the new layered image and applying the mask takes a little longer than it would to convert the image to grayscale and then playing with selection, levels and curves, but not much and I have more control over contrast and values.
You’ll find a fairly detailed guide to the last two techniques, as well as some variations, here.
For a well-exposed and full-range colour image, converting directly to grayscale and then using levels or curves will sometimes to an adequate job. Once your original image wanders from well-exposed and full-range, though, a simple conversion can require a lot of work with selections, levels and curves. Converting to grayscale from the Lightness channel in LAB colour can give you a better starting point.
All of the other methods above give you more control over the contrast and tones of your final image and are worth exploring to help you get the most out of your grayscale images. And, believe or not, there are other ways of converting images, too. Do a Google search for “color to grayscale” and you’ll find pages of possibilities.