Behind the hard process of creating Chinese computer fonts

But there are tens of thousands of Chinese characters, and a 5 x 7 box was too small to make it readable. The Chinese needed a network of 16 x 16 or more, which is at least 32 bytes of memory (256 bits) per character. If you could imagine a font with 70,000 low-resolution Chinese characters, all the memory requirements would be more than two megabytes. A font with only 8,000 of the most common Chinese characters would also need 256 kilobytes to store bitmaps. The total memory capacity of most personal computers in the early 1980s was four times greater.

As serious as these memory challenges were, in the 1970s and 1980s the most fiscal problems facing China’s low-resolution letter production were those related to aesthetics and design. Long before anyone sat down with a program like Gridmaster, most of the dry work came out of the computer, using a pen, paper, and fluid correction.

Designers have spent years trying to shape bitmaps that met low memory requirements and retained little calligraphic elegance. Lily Huan-Ming Ling (whether) and Ellen Di Giovanni were among the creators of this set of characters, hand-drawing sketches of bitmaps for specific Chinese characters or digitizing them using Gridmaster.

The main problem the designers had was the translation between two different ways of writing Chinese: a hand-drawn character, made with a pen or brush, and a bitmap glyph, made with a series of pixels arranged on two axes. The designers had to decide how (and whether) to try to recreate certain handwritten Chinese spelling features, such as input strokes, stroke reduction, and output strokes.

In the case of the letter Sinotype III, the process of designing and digitizing low-resolution Chinese bitmaps was well documented. One of the most fascinating archival sources of this era is a mesh-filled linker with hand-drawn hash marks, sketches that would later be digitized in bitmaps for thousands of Chinese characters. Each of these characters was carefully designed and, in most cases, edited by Louis Rosenblum and GARF, using a fluid correction to delete “bits” that did not agree with the editor. Above the set of initial green hash marks, a set of second red hash marks indicated the “final” draft. That’s when the data entry work began.

LOUIS ROSENBLUM COLLECTION, STANFORD UNIVERSITY LIBRARY SPECIAL COLLECTIONS

Considering the number of bitmaps the team had to design (at least 3,000 (and ideally many more) if the machine hoped to meet the needs of consumers) one might think that designers were looking for ways to streamline their work. One way they could do this, for example, would be to duplicate Chinese radicals — the basic components of a character — when they appeared in approximately the same location, size, and orientation from one character to another. For example, when producing a large number of ordinary Chinese characters with a “radical woman” (女), for example, the GARF team could create (and, in theory, should) only a standard bitmap, and then repeat within each character. it was there that this radical appeared.

However, no such mechanical decision was made as the archival materials show. In contrast, Louis Rosenblum stressed that the designers adapted each of these components — often in almost imperceptible ways — to ensure that they matched the overall character they appeared.

Bitmaps juan (娟, elegant) and mine (娩, deliver), for example — each of them has a radical woman — that radical has changed so slightly. In character juan, the middle section of the woman’s radical takes up a horizontal pixel of six pixels compared to the character’s five pixels mine. At the same time, however, the lower right curve of the woman’s radical character extends one more pixel outward. mine, and in character juan this stroke does not extend at all.

LOUIS ROSENBLUM COLLECTION, STANFORD UNIVERSITY LIBRARY SPECIAL COLLECTIONS

Throughout the font, this level of accuracy was the rule rather than the exception.

When we combine sketches of bitmap drawings against definitive shapes, we see that more changes have been made. In the draft version rounded (Luo, gather, clean), for example, the lower left stroke extends downward at a perfect 45 ° angle before entering the digitized version of a stroke. In the latest version, however, the curve has been “flattened,” starting at 45 ° and then flattened.

LOUIS ROSENBLUM COLLECTION, STANFORD UNIVERSITY LIBRARY SPECIAL COLLECTIONS

Despite the small space the designers had to work with, they had to make plenty of choices. And each of these decisions affected every other decision made for a particular character, as the addition of a pixel often changed the horizontal and vertical balance.

The unforgivable size of the network affected the work of designers in other unexpected ways. We can clearly see this in the devil’s problem of achieving symmetry. Symmetrical designs — abundant in Chinese characters — were distinguished in low-resolution areas because, according to the rules of mathematics, the creation of symmetry required spatial areas of each size. Bitmap networks with double dimensions (such as 16-by-16 grids) made symmetry impossible. GARF was able to achieve symmetry in many cases using only part of the overall grid: a 15-by-15 region within the 16-16-grid. This further reduced the usable space.

LOUIS ROSENBLUM COLLECTION, STANFORD UNIVERSITY LIBRARY SPECIAL COLLECTIONS

The story is even more complex when we start comparing bitmap fonts created by different companies or creators for different projects. Consider radical water (氵) as shown in the Sinotype III font (bottom and right), created by HC Tien (left) in front of another Chinese font (left), a psychotherapist and Chinese-American activist and experienced entrepreneur . Chinese computing in the 1970s and 1980s.

LOUIS ROSENBLUM COLLECTION, STANFORD UNIVERSITY LIBRARY SPECIAL COLLECTIONS

Although they seem as small as the previous examples, each one replaced a different decision (among thousands) that the GARF design team had to make, whether in the draft or digitization phase.

The low resolution hasn’t been “low” for a long time, of course. Advances in computing led to increasingly dense bitmaps, faster processing speeds, and lower memory costs. In today’s era of 4K resolution, on retinal displays, etc., it can be difficult to be as limited as aesthetically and technically, until they began to create Chinese bitmap fonts. But solving such problems eventually made computers, new media, and the Internet available to one-sixth of the world’s population.