Let's discuss strictly technical aspects of icon making
1. Lines, slopes
One of the most important elements of the composition is the outline of the object, i.e. solid line, which limits the object from the background. When creating normal-sized images, the designer occasionally thinks about defining objects with complementary outlines. This is unnecessary because of the scale: even low contrast objects do not blur into a one. Pixel graphics is different. Two solutions are available: either foreground and background colors have to be from different sides of the color wheel, or the object should be divided from the background by visible lines and shadows. I will discuss lines in more detail.
If we think more to the big-size images, we see that in order to highlight contours, we can use all (even the most complicated) angles, Bezier curves and edges. Anyway, the line will appear perfect thanks to anti-aliasing. When shrinking the image scale to the small icon size, situation changes greatly. When every particular pixel is vitally important and can ruin the whole appearance of the icon, anti-aliasing is just unreasonable. It comes that you have to consider the possible line slopes.
The slope you select for the line, determines the step of this line. This means that every line consists of primary elements, the combination of which defines its neatness and visual appeal.
For example, a 18-degree angle contains many tiny 3-pixel lines combined with a 1-pixel downshift:
This is how almost every line looks like. Basic element with joining and another basic element. However, not every angle makes the line look attractive and not annoying.
For example, here are 25 and 20-degree slopes:
It can be easily seen which line is more appealing. The 25-degree slope makes a line containing equal 2-pixel basic elements. The 20-degree angle has a basic element containing of three lines: 3-pixel, 3-pixel and 2-pixel. The angle is more complicated, thus, if not combined by other visual effects, it creates a feeling of untidiness: the basic elements have contrast which is too obvious to escape the human eye.
Here we can come to the first rule. The most "appropriate" angles must make as plain primary elements as possible. Thus, the ideal slopes are 0 and 90 degrees. In them, 1-pixel primary elements are joined without any shift and produce smooth lines. Less ideal are angles which form basic elements of 1, 2, 3 pixel with 1-pixel downshift. Surely, the lines will not be smooth, but the human sight will process the picture and present to the user what you meant to show to him. Also, the mentioned slopes can be considered correct (but at a stretch): angles with basic elements made of different lines shifted by more than 1 pixel.
Now we can touch another issue. In the last paragraph, I intentionally defined slopes with basic elements of 1, 2 and 3 pixels as acceptable. Such tiny lines, especially when a great amount of them is joined, appear as a single whole. But what do we see if we change the basic elements to 10 or 20 pixels?
Here is an example:
It can be obviously noticed that the line loses its integrity. It is not a continuous line anymore; it is a combination of different lines, situated near one another. Hardly any designer wants its creation to look that way. Thus, we come to another conclusion: if you use small slopes, which make long basic elements, it must be reasoned and used with maximum attention.
And, finally, the third aspect I wanted to mention about lines. I did write that the basic element should be as simple as it can.
For instance, a 25-degree angle can be drawn in the two (sure more of them exists) following ways:
In the first case the primary element is a 2-pixel line. In the other picture, it is made of three elements: 3-pixel, 2-pixel and 1-pixel. The same slope but the look is different. The complex primary element produced the line messy.
Such examples can be produced for almost any angle, so in your works, make sure to simplify as much as possible.
It can be said without an overstatement that the color is the main aspect of your work; not suiting or poorly balanced colors can ruin even the best idea. What can we say about the color in miniature graphics?
First, there is a clear rule: if the image looks bad when converted to grayscale, it means the colors that were selected are wrong. This rule is true not only for icons but for the whole graphic design.
Second, you will be unable to apply safe RGB. Of course, you may use internet colors only, but this ties you hand and foot. Gradients, shadows, blending and many other tricks will not be accessible to you.
Third, try to enlarge the portion of "plain" color. The more consistently colored parts without diffusion and jumps your composition has, the more clear it looks. Icons are too tiny and the over usage of complex effects makes the picture look messy.
Fourth, gradients. This is a wonderful tool which can change the look of any composition. Besides esthetics, gradients are an ideal way to abolish the "broken line" look (lines with too long basic elements). But be careful when applying them because too many gradients can kill the plain color, and be unable to fit into the gif palette.
Fifth, shadows and flares. All rules for using shadows and highlights in icon making are totally identical to the rules of the overall graphic design. The only ine I have to mention: draw everything yourself. Don't apply filters, make all shadows and flares in a different layer and then edit the opacity. If using filters, you rarely can predict what will it look like. It is unfavorable when you can't control the process of creating of a project. It is awful if you don't control it when drawing icons.
And, lastly, sixth. Nuancing is the main detail of pictogram design, which largely results the visual look of the composition. If the blue human eye in the icon has the size of 1 pixel, increase the color intensiveness. For instance, instead of 0, 131, 159 choose 71, 195, 242. It is indifferent that the human eye never has this color; in the first case it will look like a gray dot, but in the second example it will definitely be a blue eye.
Thoroughly control the picture, do not let the optical tricks to spoil your design. It is acceptable if something is not perfect from the technical point of view, but the whole composition must be perfect.
As an example, I drew the packet of juice standing before me. The picture has plain color, gradients with broken lines, shades and flares and nuances.
If I wrote about something other than pictogram design, this chapter wouldn't fit into a whole book, but the size of images restricts fonts too.
The size of symbols becomes the main issue as opposed to their beauty. Nearly in any case, only if not the letters are the main part of the project, the font size has to be reduced to the greatest extent possible.
In general, virtually any design company has its exclusive font with little letters. Such a font can be made in several hours. You can browse the web and make your own collection of very interesting fonts. First of all, I would recommend you to download the DS Pixel and Seventen 7Vedi webpage as well as the complete set of fonts from the Lakmus Lab website.
If you don't want a new font, little typesizes of Arial and Verdana will work. As a last resort, you can draw the necessary letters yourself.
There are not many rules there. First, characters cannot be less than 5 pixels high or less than 3 pixels wide.
Second, letters should not use anti-aliasing. It transforms the text into an unrecognizable set of pixels.
However, there are no rules without exceptions and the 5-point Arial appears rather understandable.