Virtual Graphic Artist Webzine: The Concept of Continuous Tone

The Concept of Continuous Tone

The concept of continuous tone means there is a non-broken range of tones from white to black with, in theory, every shade of gray represented. Mathematically, there is an infinite amount of tones and thus we call traditional photography continuous tone. Conventional chemical film gives us many grain particles. During exposure they are charged in a pattern that perfectly resembles the image being exposed. But each grain will not be either converted to solid black or left as white. Each grain receives a variable amount of exposure that affects some of the molecules in the grain. Thus, each grain may develop to a different level of density. Those with little exposure stay close to white. Those with a lot of exposure may be near black. All others may be anyplace in between.

The very next grain can potentially be something different. More than likely, it is going to be something close to the same but not exactly the same. If you now look at a bunch of them, a hundred or more in an area, they all seem to be varying slightly depending on the representation of the image. The grain crystals can also be slightly different shapes, some larger, smaller or, for that manner, just about any minor differences in shape is possible. These slight differences produce grains that are always a bit different and this is what we define as continuous tone. If all the grains have the same size, shape and tone (highly unlikely!), there will be no representation of an image; we will see a patch of gray. The constantly varying tones gives us our apparent image on film.

So how do you measure the density of each grain? The differences between adjacent grains are not measurable. Even if we could count the individual molecules there would always be a slight difference between them. The answer is that we simply do not bother trying to figure out how many tones we have available and assume we have "all of them"! Infinity, and that's pretty impressive.

But with digital film, we have a natural mathematical division to represent our tones. Our pixels are all of a fixed size and shape. The concept of infinity does not apply to digital because each shade of gray bouncing through our lens to our CCD will have to be rounded off and assigned to the nearest available mathematical tone. We only have a certain number of shades available and the naturalness of our photograph image must be assigned to one of these shades. The shape of the original image must be broken into a mathematically neat patter of consistent sized and spaced pixels. Digital images do not follow the natural shape of the image like conventional film images do. Thus, there isn't a true continuous representation of tones and shape of the image.

The number of digital shades

With digital film, there are three channels, one for each of the primary colors, red, green and blue. Each channel is normally divided into 256 different tones from black to white. Multiply the three together (256 X 256 X 256) equals 16.77 million different shades of color. Is this enough to declare digital images continuous tone? Further, the color model divides the range of tones for each of the primary colors by 256, which is 8 binary bits of information per channel. Each pixel contains the assignment of a tone from each of the three primary colors. But it is also possible to divide the range of a primary color channel by more than 8 bits. Ten is another possibility and will yield 1024 tones of tone per channel. Now our number of tones can be slightly over one billion different colors (1024 X 1024 X 1024). Is this enough? If not, we can have 12 bits per channel! Is 68 billion different possible tones of color for an individual pixel enough?

But how many tones can we actually see? Are our eyes sensitive enough to distinguish the difference between to adjacent colors out of 16 million? The issue of digital or conventional continuous tone should revolve around the concept of what we can and cannot see. At some point enough is enough and more tones may not make all that much difference because we may not be able to distinguish between them so why bother collecting them? Why bother making our files significantly larger with digital information that cannot be seen?

But there is another reason why we might need more tones, even if we cannot see them. A finer division of tones allows us to make better adjustments in programs such as Photoshop. This is especially true of the two ends of the spectrums. We would like to have more representation of different tones in the near highlights and near shadows. In this way we could manipulate them enhancing them as necessary. This will allow us to bring out more critical details that may otherwise by averaged out by lower number of divisions like 256. Think of the entire spectrum as 256, so how many define the near shadows? Near highlights? With more levels we have more mathematical control and that should make working with Photoshop better. Maybe we can't see the difference between the tones but Photoshop can. There will be better transitions between tones that, even though we cannot see them, can help produce pictures that are sharper and clearer.

Revising the concept of continuous tone

Digital photography forces the issue on what is and is not continuous tone. With the randomness of chemical film and so many tones available right from the very beginning of the process in the early 1800's, we never really had to deal with the issue. With digital we have to determine how many levels of tone are acceptable.

The first conclusion is that digital does not offer as many tones as chemical film so therefore it is inferior. But in reality, we have two issues to consider before we label our new process inferior. First, how many tones can we see? How many do we have to see before we decide that the number is adequate to show us a convincing representation of tones that our eyes will say is continuous? Obviously, someone is going to have to do this research and decide on a number. Is our current 8 bits for 256 tones adequate? Perhaps it is because it is somewhat difficult to distinguish between adjacent tones. Then why would we need more? Why divide by 10 bits, 1024 tones, when we can't see the difference except in Photoshop? So therefore, 8 bits may be enough except for those who think they can see the difference and they are probably the same people who think true continuous tone is the only way to go. We do have to be more objective in our assessment and not subjective!

A second concept is that of technical nature rather than visual. How many tones would we need to work with to be able to control our process adequately? We need to manipulate the tones to enhance and correct our images. If we don't have many tones, there is not much to work with. More tones give us more to work with and that may result in better reproduction of images. When manipulating these images, it is easy to see that 8 bits really doesn't give you much control over individual tones in the spectrum. It's very difficult to pull out individual tones and adjust them individually. Furthermore, the more tones we have, the finer the gradation between tones. More tones, such as 10 bits or 1024 will crate a smoother transition from black to white and this will give us the additional control we need.

For example, in the blueberry picture above, we might like to enhance the near white pixels on the not-yet-ripe berries (actually green and not washed-out white as shown) to get some near white tones that will show the berries as round and not white holes. Most of these white tones are represented within about 10 tones of near whites. An expanded scale might represent the same are with 40 tones (10 bits will yield four times as many tones). 12 bits might assign the same whites into 160 different tones! 40 or even 160 is more to work with than 10 and that might allow us to create artificial separation of tones to make our blueberries round.

A further problem is that more bits of information means more data and that means bigger files. Already with our 8 bit model (256 levels), files can be quite huge. 10 or 12 bits will render our files super huge! The storage problems will be enormous. And you do have to consider if there will really be a visual difference that might make this step from 8 to 10 worthwhile. When computers get faster and storage cheaper we may not care. But today, we're on the border. Professionals who need to control of 10 or 12 bits will have to pay the price. The rest of us can live happily with 8 bits.

The final problem is that of output. We will have to develop output devices such as printers that will create the accurate shades we could manipulate in Photoshop. There is nothing worse than producing a wonderful blueberry on screen and than have it turn back to a flat white area because the output device is only 8 bit. Even if the printer were able to produce different tones that an instrument such as a densitometer could detect, would our eyes see it? Photoshop could see round blueberries, the printer may create round blueberries, a densitometer could see round blueberries, but if our eyes can't, the blueberries are flat!

Whatever your opinions or those of professionals, the continuous tone argument is going to be with us for awhile as we rationalize the situation and develop standards we can all agree on. For now, or digital systems have enough tones to produce convincing images. But why not more? Improvements to the technology will probably bring them to us and, if it doesn't cost too much, we may not even notice. We may notice the better details and the ease of operation and for that, we can once and for all consider the argument ended.