Monday, July 23, 2012

Adobe Photoshop Lightroom Tutorial - Part XXIII - Understanding Channel Mixing to Achieve Effective Black and White Photos

Part I - Index and Introduction

Converting a photo to black and white may seem one of the easiest thing you can do with your photo editing software of choice. Unfortunately, nothing could be further than the truth, and if you don't do it correctly you may end up with dull images, very different from what you thought you'd get.

The Basic Fallacy: Zeroing the Color Saturation (Depending on the Tool You Use)

An approach I see very often to convert an image to black and white is zeroing the color saturation. What's worse, I sometimes hear theories about not-well-defined advantages of this technique. You get a black and white image, of course. But that image doesn't represent the luminance your eyes are seeing, let alone the intuitive result you think you'd get. Depending on the colors that are present in the shot and their saturation, differences can either be subtle or very deep.

What's interesting to note, as we'll see in a minute, is that most differences will be noticeable on deep blues and saturated reds. Think about: skies and some skin tones. Indeed, this doesn't look like a great deal.

Without going into the technical detailes about concepts like relative luminance or luma in colorimetric spaces, a photographer should understand that the luminance values of pure RGB colors aren't all equal to the human eye. In fact, many standards try to model the behaviour of the human eye and most photo editing programs provide us the tools we require to achieve predictable results, according to how the human eye works. Just for the sake of example, here's the transform matrix from RBG to CIE 1931, where Y is the luminance:



As you can see, red contribution to luminance is approximately 4 times greater than green's (0.17697 vs. 0.81240) and more than 16 times greater than blue's (0.17697 vs. 0.01063).

Another example can be seen in the transform matrix from RGB to the Y'UV color space:


We can see the contribution of each RGB channel to the value of Y' (luma, a gamma compressed luminance). Once again, the contribution of red is smaller than green (0.299 vs. 0.587) and bigger than blue's (0.299 vs. 0.114).

What can we infer from this? That green brings the greatest contribution to the luminance value, followed at a big distance by red, and finally by blue. In other words: given three pure RGB colors with  the same component:
  • Green will be much brighter than the others,
  • Red will be much darker than green but slightly brighter than blue,
  • Blue will be the darkest of all.


What happens, then, zeroing the RGB color saturation?

What happens is simple: the smaller the saturation, the more colors will tend to equal a specific grey value. We're not interested in knowing which grey tone, but the important thing is that each channel will then have the same luminance.

Here's a visual example, better than a thousand words: in the following images you can see the same color chart with different values of color saturation: 100%, 50%, 25% and 0%.

Color Chart - Saturation: 100%

Color Chart - Saturation: 50%

Color Chart - Saturation: 25%

Color Chart - Saturation: 0%

What's clear from this example is that zeroing the color saturation is not what you want during a conversion to black and white. In fact, to achieve balanced and realistic results, or results that at least would faithfully represent what you're eyes are seeing, you'd expect relative luminance between different RGB colors to be respected. That is, red should be a darker shade of grey than green and blue an even darker shade of green.

What can you do, then? Simple: use the tools provided by your photo editing program and forget about the saturation adjustment.

To say the truth, this also depends on the tool you use. The saturation control of some photo editors, such as Lightroom, behave differently than this and, in fact, apply a default mix when desaturating colors. Even so, even basic photo editing programs have got a "Convert to black and white" feature which will hopefully do a better job than you can do with the saturation slider. More sophisticated program may offer better tool, the "channel mixer" being the most interesting and flexible from a photographer's point of view.

Channel Mixing

As we've seen, it's necessary to "mix" RGB values with certain coefficients to simulate the behaviour of the human eye, as far as relative luminance across RGB color is concerned. That's exactly what the channel mixer can do for you, and that's why it's an omnipresent feature of good photo editing programs, although little known to many amateurs.

The channel mixer simply lets you change the coefficient with which each color contributes to the luminance. Raising a color's coefficient means that its contribution will be bigger, and thus will appear brighter. On the contrary, lowering a color's coefficient means that its contribution will be smaller, and thus will appear darker.

Adobe Lightroom 4 applies a default "mix" when converting to black and white which I found pretty acceptable. However, since the tool is there to use it, I almost always tweak the mix a little bit to achieve the results I desire. For example, I often raise the value of the red and orange channels in portraits in order to reduce speckles and imperfections and to slightly brighten the skin.

Here's the result you achieve in Lightroom 4 when desaturating the reference color chart:

Lightroom 4 - Saturation: -100

As you can see, it has done a much better job than Photoshop in this case (beware: I say "much better" from a photographer's point of view, not from a theoretical one). Now, reds and blues are darker than greens, as expected.

On the other hand, if you convert it to black and white, here's the result you get:

Lightroom 4 - Black & White - Default Mix

and here's the default mix applied by Lightroom:

Lightroom 4 - Default Black & White Mix

The result is somewhat less contrasted, and you can see how blues are slightly brighter (+9) and greens are slightly darker (-27).

The point is there's no right or wrong here: the Black & White mix is a tool you can use to fine tune your shot and is much more flexible than just zeroing the saturation. For example: if you want a darker sky, for example, just decrease the Blue contribution. if you want some brighter reds, just increase their contribution.

Conclusion

The bottom line is: the channel mixer is pretty flexible and you can use it at your advantage, for example to simulate some B&W filters in post production. For example, I often increase the Red and Orange channel in many portraits, especially those taken during the summer, to get less contrast in the subject's skin thus getting it brighter and removing many imperfections. In the following shot, for example, you can see how simulating a red filter has given the photo more contrast in the red highlights and resulting in a smoother and brighter skin:

Red and Oraange channels were increased to simulate a red filter

The same effect was used in this shot to achieve a similar result:

Red and Oraange channels were increased to simulate a red filter

In the following shot, the subject was very tanned and the shot, when converted to black and white, had a look I really didn't like. Once again, using an appropriate mix, I could deliver a more natural skin tone and get rid of all the subject's speckles too.



If you want to help me keep on writing this blog, buy your Adobe Photoshop licenses at the best price on Amazon using the links below.

No comments: