Lesson 4: Basic color manipulation

A working principle: firstly with the tool, not only with the eye

The more experienced you become in digital image editing, the more capable you become to scan an image rather with your brain and not only with your eyes. As a human being you are accustomed to seeing through your natural vision -as seeing is the most powerful of our senses- but for image editing you must develop your technical vision as well. You see an image which resides inside a computer, but a computer "sees" only numbers. So in order to effectively manipulate a digital image, you must be good at correlating numbers with optical features. To enhance this visioning ability, you need both theory and practise. The theory has already been given to you in the first Lesson and in advance you will realize that GIMP has all the tools you need to help you while working with your images. This combined experience will make you a better photographer not only at image editing (post processing) but at shooting as well.

Two types of images

We can distinguish images in two categories:

• non-continuous tone images -- clear and abrupt areas of color. Images of this kind are easy to manipulate, because different areas of the image can be easily isolated as targets to act on.
• continuous tone images -- typically photographs. Areas may not be easily isolated, since they have color in a continuum. Things get challenging. Areas in the image are difficult to separate, so we are challenged for sophisticated approaches to the separation problem. With various tools and tricks introduced in the next Lessons, we try to exploit the various numerical properties of the pixels themselves.

This is an important lesson on the critical concept of visualization of the structure of digital images -- and especially of continuous tone images where things get interesting and challenging.

The Histogram

The term histogram (from the greek ιστόγραμμα, ιστός + γραμμή, meant as correlation line) refers to a graphic that depicts the correlation of different types of numerical data. Since a digital image is stored as numbers, an image histogram can represent color distribution inside the image, in a way that offers an insight on things like:

• if the exposure was right,
• if the light was flat or with high contrast,
• hints about what kind of manipulations may yield acceptable results.

As we know every pixel of the image as you see it on screen has a color based on a set of three numbers for R, G, B, every one of these three being a brightness value in the range 0-255 (for an 8bit color depth). A basic histogram comes up when the computer scans the values of all components of all pixels and counts how many pixels correspond to these brightness levels. There can be various types of histograms, however all follow this basic logic.

Understanding an histogram is fundamental. Like a doctor who sees an X-ray and understands what is happening inside the body, a digital photographer by seeing the histogram understands the structure of a digital image. I strongly recommend that you do some homework studing some histograms of your own images, and soon you will be able to immediately visualize terms and situations like the following:

• Theoretically in a technically good image the distribution of the pixels extends to the whole x axis - the image has a full tonal range.
• If pixel distribution goes to the left, you have a low key photo.
• If pixel distribution goes to the right, you have a high key photo.
• If the histogram ends abruptly on the left - the curve does not touch the zero brightness point - , the shadows are clipped - lost detail in shadows, photo been shot underexposed.
• If the histogram ends abruptly on the right, the highlights are clipped - lost detail in highlights, photo been shot overexposed.
• If pixel distribution covers only the center, the image is flat - low contrast.
• If pixels have two tonal ranges on the left and the right, the image has a high contrast.

The technical quality of a photograph in the strict sense depends initially on the shooting conditions that have to do with the subject itself and the lighting. At post-processing, we can manipulate a photograph in certain ways, acting on the numerical data of the pixels. The histogram is a valuable tool to use, together with your eye, to affect the quality of an image. With the histogram you perceive the technical side, with the eye you perceive the optical side, and by constantly combining these two perceptions you become intelligent in digital photography.

The histogram is there to continuously follow you on any action you do, helping you to visualize the technical aspect of what you are actually doing on your image.

Histogram: linear or logarithmic

The histogram can be viewed at linear or logarithmic mode, with two little buttons up right on the dialog. By default it should come linear, as refered in the next video, and you should better leave it this way, the linear mode being a simpler mathematical depiction.

One-click-fixes

GIMP has some commands for fixing a photograph in the logic of "one click fix". They are the six commands under the submenu "Colors" > "Auto".

These commands have the disadvantage of being intended for the inexperienced user, however they can be useful sometimes. They may work on photos that are typically wrongly exposed, over- or underexposed, or with "technically abnormal" channels. To be practical, it is just one command in the end, so you have nothing to lose giving a try -- in the worst case it will be just a matter of an Undo.

These terms are very technical (and confusing even in english), so in case you are really ready for quite a lot of technical jargon, you can start from the GIMP's Help and search the web. When working with them, you are advised to have the histogram aside, so that you get used at estimating your own actions. Probably the three more useful of them are Equalise, White Balance, Normalize. If you give them a good try with many photos, you will soon realize that they may have very different results in different images. A command may well save a problematic photo or may well destroy it absolutely. Sometimes you can even get very unexpected results, that you may like or dislike at all. This holds true because the specific algorithms simply act on numerical data, with no consideration about what you imagine or want to do.

To sum up with the automatic quick fixes, as a beginner you are advised to use them at least for a while, but having the histogram aside so as to train yourself at evaluating your actions. In this way you will quickly gain some technical skill.

The three basic tools

These are the three basic tools for tonal/color manipulation. You are going to use them often and it is important to understand them in order one after the other.

"Colors" > "Brightness-Contrast..."

Suitable for beginners, easy to use - but a first step just not enough. However you must start with it, in order to understand what will come next - you just must start somehow! With this tool you manipulate the whole histogram as one compact tonal range in a linear fashion: something which is rarely enough when dealing with a usual photograph that includes a wealth of tones and colors. Advanced users find it helpful on manipulations on secondary items like channels and masks, as you will see later.

"Colors" > "Levels..."

With the Levels tool, you do the next step. Compared with the Brightness-Contrast, this tool offers two more important abilities:

1. you step down to the channel level affecting colors individually, and
2. you still consider the histogram as one tonal range, but now you can manipulate it by the three points that initially define a line geometrically: starting point, ending point, middle point.

The Levels tool is useful for a general tonal adjustment, or to quickly fix a general color cast on the channel level.

"Colors" > "Curves..."

As its name implies, the Curves Tool makes the third final step towards the control of the curves. It seems a bit complex to the beginner, but in fact it is simple, after reading the theory and watching the next video you should be able to use it effectively. You have the needed powers to manipulate the color, the brightness, the contrast, the transparency if the layer has a transparency channel. While with the Levels Tool the whole histogram is conceived as one tonal range and we pay attention on the shadows and the highlights, now with the Curves Tool we have the full power to act on any arbitrary tonal range by defining the curve at will, actually performing a sort of remapping down to the channel level.

With the possibilities that these tools offer through their dialogs' interface, you can perceive the structure of the numerical data under your images and you can practice and experiment on them. If you persist with these tools, you will soon be able to perform effective manipulations on tone and color. I recommend that you scan the GIMP's help on these tools, and do just that one thing: use them.