B&W Conversions - Methods and myths

[ciaran]

More tutorials can be found on my website: theWonderOfLight Dot Com

I hope this is somewhat of a definitive guide to B&W conversion. There are just so many ways of converting a colour image to B&W, each one having it's own advantages and disadvantages. One method may favour one style of shot, another favour a different type. The point is, use what works for you. There is no right way or wrong, best way or crap way and anyone that tries to convince you is completely wrong.

Now in saying that, some methods are possibly more flexible than others, so people may believe they're better and more powerful. Others may argue that because they're more flexible, they're also more inconsistent and as such they're not as good. At the end of the day, you need to decide what works for you and the particular image you're working on. Don't let any one person tell you that their method is better. Experiment for yourself, make up your own mind. Definitely don't believe the hype :!:

Rather than detailing the "how to" of each method, I've displayed a diagram below of the before and after of each conversion method, as well as including a histogram. The before shot, shows 3 pure colours of Red, Green and Blue along the top and then a colour picture below it, containing a combination of these colours. The after shot, is that colour shot, converted using the method discussed. The histogram is included mainly to show and prove differences between techniques.

Regardless of how we get there, each method produces an image which contains only B&W tones ranging from white through to black. Whilst it's possible to tone monochrome prints, for simplicity of discussion, I haven't included it here. Also, if you're interested, I explain a bit of the maths of B&W conversions at the bottom, but for now I'd like to start off easy. So let's begin..

Desaturate

Desaturating the colours is probably one of the simplest ways of converting a colour image to B&W. The myth about desaturate is that it throws away colour information. Whilst this is true, SO DOES EVERY OTHER CONVERSION METHOD, after all, we want to create a monochrome image. However the problem with desaturating an image is that it maps all colour tones to the same greyscale value, often leading to a very flat image. If you look at the diagram below, note how the red, green and blue elements all become the same tone, i.e. 33% red, 33% green and 33% blue.



Grayscale

Converting an image to grayscale is another very quick and easy method to create a monochrome image. If we look at the output converted image, it can be clearly seen that the red, green and blue channels all get mapped to different tones, leading to a different black and white image. This DOES NOT MAKE IT BETTER OR WORSE... JUST DIFFERENT. Certainly, I'd prefer to use gray scale than desaturate in general, but if my original colour image comprised mainly of one dominant colour, then the output converted image would be more or less the same regardless. Converting an image to gray scale, once again throws away all colour information. Generally when we convert a colour picture to gray scale, we take 30% of the red value, 59% of the green and 11% of the blue, to combine to give us the final output. The values are chosen due to the different relative sensitivity of the human eye to each of the primary colours. Photoshop does this automatically, but try entering these values into Channel Mixer and see for yourself.



Luminosity/Lightness

Most people think of each colour as a combination of red, green and blue. This is an RGB colour mode. Another common colour mode is CMYK which is commonly used by printers (Cyan, Magenta, Yellow, Key/Black). Another yet again is LAB (Lightness/Lumiosity, A and B). The colour is split into the A & B channels where Lightness contains the luminosity/brightness of the pixels. Converting via LAB mode, uses this luminosity mode as the base of the B&W conversion. It is colour agnostic in that the colour information is ignored for the conversion and only the relative brightness of the pixel is considered. Unlike the previous two modes, this does not have a channel mix equivalent, because the colour information is not considered before the conversion. So there is a VERY distinct different between it and converting via grey scale, as converting via grey scale definitely favours one colour over another. The differences are subtle for the below example image, but can be seen in the histogram between both images. Similarly an image with a huge cast of one colour over another will convert differently depending on what method is used. When people shoot B&W film, assuming no filter is used, the film is colour agnostic, so converting via this method should give you film-realistic results.



Channel Mixer

Channel mixer is a wonderfully powerful and flexible way of converting colour images to B&W. This is because we as the use can control the mix of Red, Green and Blue values, which make up the final gray scale image. If we use 100% red, then red will become 100% white and the green and blue components turn completely black.



If we choose 100% green, then the green element goes to pure white and both red and blue turn 100% black.



And similarly, if we choose 100% blue, the blue channel goes completely white and both the red and green channels turn black



The beauty of channel mixer, is that we can change the mix of channels to achieve a different B&W conversion. This is usually achieved as an adjustment layer, so the original colour information is untouched. HOWEVER this does not mean, we are not throwing away colour, because we are :!: All we are doing is controlling how much each of the individual 3 channels contribute to the final gray scale brightness. If we changed the R, G, B values to 33, 33 and 33, we'd effectively have a desaturated image. If we change them to 30, 59 and 11, we'd effectively have grey scale.

Gradient Map

Without going into the explanation of how these methods work, there are numerous other ways to convert a colour image to gray scale. One such method is via a gradient map, another would be via calculations. Regardless of how they work, you still end up with a B&W image.



Summary

So each method has it's uses, it's advantages and disadvantages. Some favour channel mixer for it's flexibility in being able to change the mix of colours lending to the final conversion and it's non destructive nature. Some favour LAB for it's "film like" result and consistent approach in that it is colour agnostic. Other's favour gray scale or desaturate for their pure ease of use. The point is find what's right for you. Or better still, find what is right for your current image. Don't believe anyone telling you one method is crap or one is better than another.. it's simply not true. The best thing you can do when you hear this is ask them to prove it

The Maths

As with all explanations of any concept, it's necessary to make certain assumptions. So to ease this explanation, let's assume for a second we're in a 24 bit RGB colour mode, often referred to as true colour. So what does this mean? Well as mentioned above, we have a red, green and blue component for each colour. The combination of all 3 gives us our final colour for the pixel. Each colour is made up of 8 bits (a string, 8 long containing any combination of 1's and 0's). This is binary code, and the minimum possible value is 0 (00000000) and the maximum is 255, giving a possible of 256 values. We then combine all 3 colour values, to give us our final pixel value. So for instance in the colours above, red, would be made up of 255,0,0. Green would be made up of 0,255,0 and blue made up of 0,0,255. For colours in between, we now have a possible 16.7million colours or 256x256x256. This is obviously massive :!: By the way, the reason it's called 24 bit, is that we have 8 bits for red, plus 8 bits for green, plus 8 bits for blue.

So.. what about B&W? Well I'm afraid to say, what we are doing is dumping 2 of those channels... NO MATTER WHAT MODE IS USED :!::!::!::!:. We may have 16.7 million colours, but unfortunately we only have 256 shades of gray. Let's just take channel mixer for instance, which some people seem to favour. Look at the tick box when you tick "Monochrome". What do you see along the top?? Yup.. the answer is "Gray channel output". All the channel mixer is doing is telling you what combination of what colour channel makes up the final gray scale image. There is only one channel :!: So why do we still have an RGB image? Well because most applications/printers/browsers like RGB colour spaces. But once converted, all channels have identical information in each. But the channels/shades of gray are the same. So, contrary to the belief of some, when converted, no matter how we got there and assuming no toning, we really do only have 256 shades of gray.

In the end, you need to forget the lingo, forget what one person tells you over the other. Find what works for you and your image. Always listen to advice, but don't necessarily take it as gospel. Not even this.

[Lee4145]

Good explanations. Thank you for taking the time.

but don't necessarily take it as gospel. Not even this.

And karma for this chuckle.

[Thunder9]

Very informative and well written

[Rense]

good write up.
Thanks for sharing.

[ciaran]

Thanks everyone

Originally Posted by Lee4145

Good explanations. Thank you for taking the time.

And karma for this chuckle.

[Dean Johnson]

Very well put together

[Mr. Pickles]

Pretty similar to this post....

B&W Conversion Methods

[ciaran]

I think the focus is different. That thread focuses on how you do the conversions, where as I try to describe the difference between each of the conversions, with examples of before and after pictures, as well as including histograms.

My tutorial was written to disprove the myth that one conversion method was THE BEST above all others. Each one has it's uses, advantages and disadvantages.

I apologised in one of my tutorial posts for the possible duplication of information, but the reason I posted them was because I was asked to by the admin.

Originally Posted by Mr. Pickles

Pretty similar to this post....

B&W Conversion Methods

[jfrancho]

I think a lot of this will fall to the wayside, as more and more use Lightroom to do the conversions, using the new B&W controls in raw conversion, and the advanced B&W dialog in CS3. Personally, nothing beats getting a good exposure to build a good digital B&W image from. Also, the conversion method seems to be the least important of all the steps in making an effective digital B&W. Personally, the channel mixer and the LxL method Rense describes works for pretty much any well exposed image. A few things to remember about CM conversions: green channel is great for skin tones, red for rocky landscapes, and blue can be quite dramatic in scenics with people. Really, it isn't so much about what method is better, as you say, but what's easiest and allows the most tuning. Speaking of tuning, The big difference in an digital B&W that makes or breaks is is in the finishing touches: localized contrast, burn and dodge with an overlay, using all the 256 tones, where the predominant tones lie, sharpening, blurring etc. martin Henson has a pile of good information that shares a few of his and others' secrets: digital-monochrome fine art black and white landscape print galleries.

There is a cool graphic here that aligns the CM with the results to show how the final product can be tuned: Digital Black and White in Photoshop - Photo Tips @ Earthbound Light

Finally, one of my favorite resources, TLR. There is a ton of good info here, including fine art B&W, toning, sharpening, didge and burn, etc.: Welcome

[gpf]

Fantastic thread. Thanks.

[Paul Shields]

Fantastic thread ciaran. Keep them coming. You're adding a lot of great resources to the camel.

[Paul Shields]

Originally Posted by ciaran

I think the focus is different. That thread focuses on how you do the conversions, where as I try to describe the difference between each of the conversions, with examples of before and after pictures, as well as including histograms.

My tutorial was written to disprove the myth that one conversion method was THE BEST above all others. Each one has it's uses, advantages and disadvantages.

I apologised in one of my tutorial posts for the possible duplication of information, but the reason I posted them was because I was asked to by the admin.

Cairan - no need to apologise. This is an excellent tutorial. There may well be other excellent tutorials posted here on the camel, but I don't see this encroaching on any others in any way.