Flexible and intelligent typographic systems

Flexible and intelligent typographic systems Presentation of article Daniel Rodríguez Valero Department of design & Image Faculty of Arts University o...
Author: Lindsey Bond
2 downloads 0 Views 142KB Size
Flexible and intelligent typographic systems Presentation of article Daniel Rodríguez Valero Department of design & Image Faculty of Arts University of Barcelona Pau Gargallo, 4 08028 Barcelona [email protected] The uncertain days of font wars are over. Now we have powerful computers, systems for visualization and printing at different resolutions, and all kind of applications for entering and formatting any text. Also we have access to subtle typographic features, which can be programmable and context sensible. All these can be sent troughout internet in numerous idioms and multi-platform files. ABSTRACT.

But maybe we lost something on the road; it’s time to review the work done by the great pioneers of digital typography and to apply their great ideas to our material. The standard font formats (PostScript, TrueType and OpenType) are too rigid for the user and the alphabet designer, so in this paper we’ll check the flexible and intelligent typographic systems like METAFONT, the experiments of the late 80’s (Jacques André, Letterror, Hermann Zapf, among others) and the interesting developments of mid 90’s (Ares Chameleon, Incubator, Intellifont) which unfortunately disappeared under the pressure of the market. Any analysis process requires also some synthesis capabilities, so we’ll present some advances in the creation of a new system called Constructor, suited for the construction of parametric alphabets directly on PostScript language, based on calligraphic knowledge similar to that used in METAFONT. KEY WORDS:

font creation, calligraphy, parametric fonts, intelligent fonts

Title of the journal. Volume X – no X/2002, pages 1 to n

2

Title of the journal. Volume X – no X/2002

Introduction Now, 26 years after PostScript®, type design is a matter of drawing outlines in programs like FontLab® or Fontographer®. The main reason is that standard font formats are outline-based: PostScript Type1, TrueType and OpenType work with quadratic or cubic splines, connected to draw the limits of black and white for each character in a font. There are two concepts underlying this argument: each letter has always the same appearance and the best way to generate outlines is drawing them directly. The former is a cultural legacy from the previous industrial technologies, based in the production of a mould and the reproduction of identical copies of the image carried in it. When computers were adopted as the main tool for font production, this assumption still survived, and each character is printed and displayed as an identic copy of the original. This doesn’t respect the inherent logical of the system, which is more elastic. As Jay D. Bolter wrote: “Change is the rule in the computer, stability the exception, (…)”1 The later is supported in another technological tradition: in 1885, Linn Boyd Benton completed the development of the first pantographic punchcutter, a machine which was able to engrave the matrix and scaling a single font design pattern to a variety of sizes. Before Benton some Renaissance artists (Feliciano, Pacioli, Palatino, Durero & Tory, among others) tried to find the geometrical construction of roman capital letters drawing their limits. The same method was used by Sébastien Truchet to design the Romain do Roi (1695) and also by the lustre engravers to make jewells and coins: outlines first, filling afterwards. The use of the pantographer in the font production carried another problem: the same drawing is used for all sizes. This represents a great break with the tradition of punchcutters, who made slightly variations in form and proportions on each body size, in orther to prevent the perceptive changes produced by the scaling of the letters. Let’s have a look to the response given to these problems by a few and interesting researchers:

1. Dynamic fonts The first dynamic font was created by Donald Knuth in 1980, first published in 19882. It was based in punk graffitys of the 70’s, so each stroke was distorted by random functions, as if

1 2

Bolter, Jay David. Writing Space. Nueva Jersey: LEA Inc., 2001, p. 5. Knuth, Donald E. «A Punk Meta-Font». TUGboat, 1988, vol. 19, núm. 2, p. 152-168.

Short title of article

3

they were painted in a rough wall. Jacques André translated it to PostScript in 19893 to make it more dynamic, and adding random placement for each point of the stroke. Letterror4, in the same year, released Beowolf (FontFont), another random font with an special function called “freakto”, based on “lineto” and “curveto”, which could specificate the location of each point of the contour of the letter inside a given space. In 1993 they tried another possibility with their font Kosmik, drawing three different glyphs for each character that will be called for in the right order (they founded only one word with three letters: Sauerstoffflasche). As they pointed in an article for the Magazine Émigré:5 “Through our experience with traditional typesetting methods, we have come to expect that the individual letterforms of a particular typeface should always look the same. This notion is the result of a technical process, not the other way round. However, there is no technical reason for making a digital letter the same every time it is printed. It is possible to calculate every point and every curve differently each time the letter is generated bby slightly moving the points that define a character in various "random" directions.(…)” These are interesting experiments, showing us the force and the intelligence of the machines we are using. Why should be go on having an industrial mentality in the era of desktop publishing? We can prepare unic & single documents with our own equipement, acting more like craftmans, personalizing our material. The market has probably the last word, because it has to move in another direction: now foundrys sell their fonts without any kind of dynamic behaviour.

2. Drawing contours The main aplications to build up fonts are based in this method: drawing directly the contours that will be stored in vector formats (PS, TT or OT). Ikarus (since 1974), Fontographer (1985), FontLab (1992), DTL FontMaster (around 1995) and FontForge (2000) are the most popular programs to design typefaces: all of them work this way. But the need for storing outlines doesn’t imply to draw them directly; in fact, screens and printers need to work with bitmaps and we do not use bitmap fonts: we rasterize outlines and render them, so vector fonts are a medium to get some result. We could follow this principle and create a system with intelligence to draw itself the outlines of the characters of any given font, as Donald Knuth did in the 70’s to develop METAFONT. He used a calligraphic way of thinking to explain the machine how to draw bitmapped letters. Two of their best students created the extensions to create outlines: John Hobby’s METAPOST and Richard Kinch’s METAFOG. METAFONT is difficult to handle to standard users (it has 62 different parametes and requires programing knowledge), so not many alphabets have been created in this particular way, but it’s an important lesson for researchers, an open-minded solution for a complex problem.

3

André, Jacques. «PUNK: de Metafont à PostScript». Cahiers GUTenberg, diciembre de 1989, núm. 4, p. 23-28. 4 Middendorp, Jan. Dutch Type. Roterdam: 010 Publishers, 2004. p. 209-218 5

Available in http://www.letterror.com/content/best/index.html

4

Title of the journal. Volume X – no X/2002

3. Optical scaling

METAFONT is the first system with so-called optical scaling6. Each body size is rendered independently, so variations can be made to the parameters that control the final shape. Jacques André worked in non-linear scaling inside PostScript, writing some interesting papers on the field.7 Peter Karow, Ikarus developer, programed KQ-program in 1991 inside HZprogram, in colaboration with Hermann Zapf: the interpolation was the main tool. In 1992, six years after the final version of METAFONT, Adobe released Multiple Master, a system working with interpolation between masters in one or more axes, which finally was able to produce a PostScript Type1 font with the instance created by the user. This fonts were expensive and no design aplication permited to create instances on the fly but Illustrator. Adobe discontinued the development of Multiple Masters in 2000 as part of a deal with Microsoft that gave us the OpenType format as a result. In the middle of the 90’s there were also some nice programs like Ares Chameleon and Incubator—both have desapeared—which offered this feature, and we should also mention here TrueType GX, the alternative to Multiple Masters from Apple, which unfortunately didn’t get the support from the design aplication’s manufacturers. Now we have OpenType, and there’s a feature called fontsize which—we hope—will be ready in some font families in orther to tell the composer which font is adecuate to the related body size.

4. Programing and designing Ther’s one fact designers can not reject: we have to learn how to program. Main design aplications combine direct drawing and composition with program languages to avoid repetitive tasks and create new effects and tools, such as Flash with ActionScript, InDesign with JavaScript or FontLab with Python. Let’s not use Photoshop filters, let’s create new ones. When the program can not do what the project needs, it’s time to develop new programs or new scripts inside the aplications. This aim has moved us to create a new system to produce alphabets, easy to use, with a few number of parameters and inspired in calligraphic notions.

6 We prefer to say perceptive or non-linear scaling, because optical scaling can be confused with photographic scaling. 7 André, Jacques y Delorme, Christian. «Le Delorme un caractére modulaire et dependant du contexte». Communication et langages, 4º trimestre de 1990, núm. 86, pág. 65-78. André, Jacques y Kinchin, Corinna. «Adapting character shape to point size». PostScript Review, mayo de 1991, vol. 3, núm. 3, pág. 31-36. André, Jacques y Vatton, Irène. «Dynamic optical scaling and variable-sized characters». Electronic Publishing, Diciembre de 1994, vol. 7, núm. 4, pág. 231250.

Short title of article

5

5. Our proposal In the mind of a drawing-like method, these two letters are correct:

Both contours have the same number of control points, the direction of the outlines are ok, so there’s nothing wrong with them. But only the “A” on the right is correct for our eyes, why?. One can think in historical reasons, tradition, customs or habits, but the fact is that the left stroke is thiner than the right one because of the tool used for the creation of that form: the reed.8 This is the main tool in western writing, while the brush is the one at the east (excluding indian, semitic and arabic writing). The reed or the broad nib (they share the same principle) create different strokes depending on the direction and the angle of the front. In roman letters, the angle uses to be from 20º to 30º: when it moves down to the left it creates a thin stroke, when does to the right a thick one. The next diagram shows this relationship:

Thus a deep knowledge about calligraphy can be very useful to design typefaces. Although they are two different kinds of writing, they share a lot of structural and formal elements: the letters of our alphabet, whatever they are hand-writed (chirography), drawed or engraved (lettering) or prefabricated (typography) have a common skeleton: we call it cursus or course of the letter, the path of the tool over the writing surface. This tool has a certain shape and a behaviour during the traslation, which determinate the final form of the character. We call that modus or modulation. Big changes on the direction of the cursus of the letters create new 8 Catich, Edward M. The Origin of the Serif. Brush Writing & Roman Letters. St. Amvrose College, Davenport, Iowa: Catfish Press, 1968.

6

Title of the journal. Volume X – no X/2002

alphabets, changes in their modus generate new components of the graphic family (bold or light, for example). The sequence and order of strokes (ductus) is determinated by the tool. Changes in cursus are discretionary, but changes in modus are systematic, so they can be traslated to parameters. Let’s have a deeper look to the behavior of the cursus. There are two kind of strokes, depending on the direction of the front: downstrokes, which are sliding and easy to execute, and upstrokes, which are pushing and difficult. The letters made with downstrokes are named roman, they are interrumped and intransitive. In the opposite direction we have cursive letters, made with down and upstrokes linked together, they are continous and transitive.

In the other hand, we have the modus, determinated by the shape of the tool and the movements during the path. The tools used in western writing are basically two: the broad nib (an evolution of the reed) and the pointed nib (used by the first time with gothic cursives around the eleventh century and a standard for the Ranaissance men). The broad nibis rigid, produces regular strokes, centered to the cursus, we call that traslation. The pointed nib is flexible and thin, and it produces varuiable strokes, depending on the pressure of the hand. We call this expansion.

A good calligrapher can also make variations on the angle of the nib during the movement of traslation or expansion. We call that rotation of the front. The combination of these elements

Short title of article

7

gave the writers a greate number of possibilities, so let’s make a brief history of writing in a few lines: 1. Antiquity and Middle Ages: after the invention of the alphabet, the reed and the broad nib were the main tool for soft text production. The parameters used were traslation+no rotation.

2. Renaissance: flexible pointed nib, expansion+no rotation

3. Manierism: broad or pointed nib, traslation or expansion+rotation

4. Romantic and Neoclassic: flexible pointed nib+no rotation

5. Contemporary: flexible tool (brush), expansion+rotation

8

Title of the journal. Volume X – no X/2002

We understand typography as writing with prefabricated letters9, so no matter if those letters are influenced by calligraphy or lettering. At this moment calligraphy has no or little influence in typography, because modern technology (vector graphics) works the same way that lettering does, this is, with stroke sequences. We want to say something to this respect, and we are not alone: This particular way of thinking took us to a new chalenge: We wanted an easy-to-use system, so wen we found METAFONT it was discarded…

5.1. Examples The first step is to draw the cursus in EPS format, the aplication creates a sequence of vectors with a behavoiur controled by a few parameters (thickness, kind of contrast, rotation, and centerness) and finally makes the outline for editing if necessary.

There are some general parameters governing some important aspects, like horizontal scale:

We can also change the global thickness:

9

This definition comes from Gerrit Noordzij

Short title of article

That means that is possible to make a complet font family in a few minutes:

We can also change the contrast:

The kind of construction:

9

10

Title of the journal. Volume X – no X/2002

Or we can ask the aplication to choose the values of the parameters inside a rank given by the user:

This last option is very interesting, because we are now working with a new concept, more flexible, adapted to the machine, not imitating old forms made with old machines. Another tests have been made to prove if it’s good to make digital revivals of old lead material, and we concluded that it’s difficult to make detailed subtilities. The easy of use of the system permits to combine both worlds, calligraphy and lettering. Constructor, our program, can be used to determine the structure and the main form of the letter, and final details can be easily done with vector tools. Changes are fast and easy, and there’s no need to redraw each letter everytime to make bold or condensed versions. Constructor has been made specially for my PHD Thesis, in collaboration with Marc Antoni Malagarriga I Picas, who wrote the PostScript code. Once we tested and we are sure that it works properly, the next step is to produce a user interface. The first version is a PHP module for remote execution (special thanks to Vicente Boix Díez for his help with the concept) made in collaboration with Alquimia da Cor, a company from Porto, Portugal, and we are now looking for partners to collaborate in a second interface in a high-level programation language to make a local execution module, suited for Linux, Macintosh and Windows OS. We want to register it as a part of the Open Source project, alas we want everybody to use our system with no limitation.