Notes from a slave to Multiple Masters

From BenningtonWiki

From: Joe Holt (6/2/92)

Notes from a slave to Multiple Masters

As some of you may know, I am one of the engineers exploring what Adobe Illustrator’s future might be. I’ve spent a lot of my time recently talking with artists in and out of Adobe, trying to get a feel for what they do, and how our program can help them. I didn’t intend to get sidetracked with Multiple Master fonts, but I discovered that MM introduces many new concepts to typography for which there are no precedent, and so consequently there are no user models to support them. I decided to try some ideas, run them past people, make changes, run it past ‘em again, and see what we end up with. I wasn’t planning on making any concrete decisions about MM use in applications, but just to try out some possibilities.

Headline copyfitting was one such possibility, and that’s what I concentrated on for much of the prototyping. The program is called “Headlines.” It tests an approach to directly manipulating MM fonts in real time for a specific task: fitting a line of text within fixed margins. It also explores methods of selecting MM fonts from a font menu and how to display MM fonts if their numbers have been tweaked by an algorithm. Some of you have seen this and I appreciate your time. I’d love to give a demo and hear feedback from anyone else who wants to see it.

I think I learned two main things from writing and testing Headlines. The first is that there are “second-order” uses of MM which are closer to what people really want to do, as opposed to giving users font “knobs” and letting them create instances left and right; and second, there are good ways and bad ways to select MM fonts.

To understand my program, it may help to look at it in the entire context of MM usage (as I understand it). What do illustrators do that the flexibility of MM could improve? Here are a couple of my ideas:

1. display text effects,
2. display text copyfitting,
3. body text copyfitting, and
4. font substitution.

1. Display text effects are things like blends and ramps from one MM instance to another, or picking an instance in the design space that is “exactly” the right width or weight—being able to pick something that’s just the right color for the title page of a book or article, for example. This use is the first impulse of an application writer wanting to incorporate MM: give the user a palette full of sliders and let him or her pick any instance in the design space. This is a very exciting use, but I think that this exposes too much of the technology right up front, putting all of the knobs and dials in front of the user who is unfamiliar with MM. It’s kinda like the old synthesizers with a rat’s nest of patch cords and knobs everywhere. Definitely fun, but not for the weak-hearted.

Of course picking a specific instance of a MM font isn’t just for display text. I believe entire books will be set in unique instances of our fonts. This use—in effect having a type designer create a unique font for you for a specific work—is very exciting. For the purposes here, a user-created font can be thought of as another primary or blessed font.

2. Display text copyfitting is what Headlines is. Let the user pick a font, then change it a bit to fit within a set of constraints. In this case the constraint is line width. MM’s flexibility is used as an extra “plus”— instead of just altering word or character spacing, a program can alter one or more MM axes to fit the text. This can give decent to very good results, depending on the application. “Headlines” just alters the MM width axis, but a real application should allow altering character and word spacing as well as point size, all in an attempt to fit the text within a fixed line width. Notice that in this use, the artist is removed from having to know the “knob settings” of the MM font: he or she picks Minion “Regular”, and the numbers are changed from that to fit the text.

3. Body text copyfitting is a more complex and subtle use of MM. Today, paragraphs are fully justified by altering the space between words, and sometimes the space between characters. When a line is shorter than the margin, the extra space is distributed over the entire line to stretch it out. If the line is a bit too long, space can be removed throughout the line to get it to fit. Both of these methods alter the amount of white space in the line, and consequently the ratio of white to black, or color, of the line changes. Now if the paragraph were set in a MM font, the characters on a line could themselves be made slightly wider or narrower to assist in fitting the line. The changes would be small enough to each character that it would only be noticed on very close examination. But nonetheless it would be enough to reduce visually the amount of white space added to or removed from the paragraph, and so the paragraph would be closer to its optimal color.

If the line breaking algorithm knows about this flexibility, it may be able to break the lines more efficiently, chosing to set a paragraph differently in order to avoid hyphenation. So MM can improve both the color and setting of a paragraph.

Again notice that the user is removed from knowing the specifics of the MM manipulation. It happens behind the scenes. The user picks an instance (probably one of the blessed or primary instances) and the algorithm adjusts one or more axes very slightly to get the play that it needs.

Unfortunately there haven’t been any real studies done to determine if you can get enough play to make a difference, without seeing the difference. Interestingly, the type people upstairs are split on this issue.

4. Font substitution might be useful to artists, but I suspect it will be more useful for document interchange—a larger concept, but one from which artists would nevertheless benefit.

All of these are serious possibilities and need study to see how illustrators and designers could use them. I think Multiple Master fonts are really a revolution, and we’ll see things we haven’t dreamed of as artists and others get their hands on them. It’s our job to give them all of the flexibility of MM fonts integrated into how a designer works. It’s very important that we provide some light on this, bearing in mind that our most ambitious efforts will be just the beginning of MM font usage.

The other issue is in the interface. Really, MM fonts don’t exist until you’ve chosen a spot in the design space. Any spot will do, and this is almost universally overwhelming at first. To guide the user along, the designers have chosen “points of interest” in the design space, such as “Regular” and “Bold Condensed”. It’s as though the type designer were standing there in front of a map, saying “You can go anywhere you want, but here’s a spot that you might like.”

Unfortunately, the points of interest aren’t called “Regular” and “Bold Condensed” but “400 RG 600 NO” and “700 BD 300 CN”. However, I don’t see any problem with that—I don’t live in Mountain View, I live near 37.26 LA 122.5 LO.

Here’s a quiz. You should do pretty well, working at Adobe and all. Match the MM styles with the common names:

1) 215 LT 700 SE		a) Semibold
2) 400 RG 300 CN		b) Light Semi Extended
3) 565 SB 600 NO		c) Black
4) 830 BL 600 NO		d) Condensed

Which column did your eye go to first?

My point is that we have two contradictory intentions at work here: one is to show users that there is nearly an infinite number of possibilities in any one MM font, while at the same time holding their hand and not losing them in the expanse. In other words, we almost need to say, “You can go anywhere you want, but you’ll probably end up here or here.”

To compound things, imagine thinking of fonts by number, as in the first column above. I choose “565 SB 600 NO.” I use this font in copyfitting, and the algorithm has decided to alter the width a bit: the 600 is changed to 597. The question is: what font is now checked in the font menu? Is it still the font I chose, “565 SB 600 NO,” even though it’s now “565 wt 597 wd”? Or does a new entry appear with the tweaked values, changing as the algorithm decides upon different adjustments? What if I select the text and paste it with text attributes into another text block. What numbers are used there?

Obviously users want the original font they’ve chosen to stay associated with the text, regardless of changes made by the application. These small changes can be likened to kerning or word spacing due to justification—attributes that change with context, but don’t affect (in the user’s mind) the actual parameters of the text. So the need becomes obvious for a style name that the user can associate with the text, regardless of changes made to the actual numbers.

I’ve found a very interesting learning progression as regards MM style names. All new users prefer the traditional typographic names. The numbers are meaningless to them. They want to pick Regular, Italic and Bold, and they want to see them that way in the menu. However—and this is what’s interesting—after using MM fonts for a while, and after becoming familiar with the numbering convention, the numbers begin to make sense to them, and in fact they become a new language, a new way to think about fonts. They begin to see the continuum upon which the different styles lie. And so they begin to prefer the numbers. One designer told me that she can close her eyes and imagine the set of numbers she needs for a particular look! What was once meaningless has become significant and empowering.

So I believe at this very early point in the exploration of MM font interfaces that we must support this learning progression. Give the styles common names, but leave enough of the numbers in the interface that the artist can clue in to their significance, and eventually learn the new language and become capable of using it and thinking in it.

(This reminds me of street addresses in Utah. It confuses me, but residents swear by it.)

The approach I use in Headlines to reflect small changes to a chosen font is to append those changes to the font name as small positive or negative numbers. For example, the “565 SB 600 NO” example above works like this: the user picks “Myriad Semibold.” If the algorithm changes the 600 to 597, “Myriad Semibold” is still checked in the menu with no name change. But in all places where the name is displayed for feedback, it is displayed as “Myriad Semibold -3 width.” This maintains the original identity chosen, but reflects that changes have been made. An alternative would be to display the absolute width, as “Myriad Semibold 597 width.” This increases the exposure to the real numbers behind the names. You can imagine how this might be done graphically.

Renaming the MM styles is just one aspect of the interface. There are other teasers that I think will only be resolved over time and after not a few mistakes. For example, how will people interact in real time with MM fonts? What uses of MM fonts should be invisible (copyfitting and widow control, for example)? How invisible should “invisible” be?

I’ve spent more time than I imagined working on this. I’ve found Multiple Master fonts to be very exciting; the possibilities are intriguing. The one impression that keeps recurring is that there is a lot of exploration for us to do, thinking about uses and coding prototypes to get the ideas into the harsh reality of users. How are we going to promote our own technology? How are people going to be using MM fonts in the coming years?

That’s where I’m at right now. If you have anything to add, please let me know. What we explore right now will directly influence the way artists and designers work with type on their computers. Makes me dizzy thinking about it!

P.S. Answers to the quiz: 1b, 2d, 3a, 4c. Did it help that the MM styles were in alphabetical order?