Demystifying the birth process of the first batch of Chinese computer fonts, how difficult is it to “move” Chinese characters into digital devices?

When a new electronic device is turned on, the first interactive page shown on the screen is often the system language choice.

You can swipe up and down to select hundreds of characters such as Chinese, English, and Japanese. But if you set the time back more than 40 years ago, this option may only be a few, and there is no Chinese.

The history of Chinese writing has a long history, but its digitization process is not long. The reason why we are now able to read Chinese on electronic devices is inseparable from the group of people who initially spent a lot of effort to "move" Chinese onto the computer.

Recently, Stanford University has acquired more than 2500 modern Chinese information technology collections, including dozens of rare Chinese typewriters, word processors, and computers. It can be called the world's largest collection of modern Chinese IT history.

▲The first batch of Chinese digital font models. Picture from: Stanford University

Thomas Mullaney, a professor of Chinese history at the school, found many interesting stories in these precious collections. Including how the world's first batch of Chinese digital fonts were made.

Thomas described this difficult but artistic journey in " MIT Technology Review ". We also have the opportunity to see this historical fragment of epochal significance.

▲Thomas Mulani

The opportunity brought by a machine

The story starts with an order.

In the early 1980s, the Graphics Arts Research Foundation (Graphics Arts Research Foundation) found Louis Rosenblum and wanted to invite his team to create a Chinese font for the machine Sinotype III under development. .

At that time, Louis was nearly 60 years old, and he graduated from the Massachusetts Institute of Technology. He was a senior printing and typesetting expert. Louis founded Photography Systems in 1965, specializing in digital engineering, photography, applied mathematics and other related problems.

▲Louis Rosenblum

Although Louis and his team had worked with the Graphic Arts Research Foundation many times before, the project of creating Chinese fonts for Sinotype III was the most difficult one.

Because China hadn't started to produce personal computers at that time, computers produced in other countries or regions could not handle Chinese. Therefore, before developing Chinese fonts for the experimental machine of Sinotype III, Louis's team needs to program the Apple II computer so that it can run in Chinese.

▲Apple II. Picture from: Wiki

All beginnings are hard. Since Apple's second-generation DOS 3.3 operating system cannot input and output Chinese character text, it must be programmed from scratch, including writing a Chinese word processor. To this end, its team spent several months of effort.

The solution they came up with was to first write a "Gridmater" program through the BASIC programming language, and then put the program on the floppy disk of Apple's second-generation computer to run. In this way, digital bitmaps of Chinese characters can be created and saved.

Then, insert the designed Chinese character bitmap and its corresponding code into the system database, and the Sinotype III machine can process and display Chinese.

▲A photo of the Sinotype III monitor, showing the Gridmaster program and the Chinese character "電". Picture from: Stanford University

Insert a background knowledge here. Early digital fonts were displayed using bitmap images (also called bitmap images).

This is a common way to store images. The photos and screenshots taken by our camera today are all bitmaps. A picture in JPEG, BMP, GIF and other formats is composed of many pixels, which are arranged and dyed to form a pattern.

For example, we can zoom in on a picture on the computer, zoom in to a certain extent, and then we can see the square pixels. Early fonts were formed by arranging and dyeing in a grid of a certain size.

▲ Zoom in on the eyes on the left picture to see the pixels

Chinese digital fonts are much harder than English

The first reason why Chinese digital fonts are difficult to make is that there are too many Chinese characters, and the second is that the glyphs of Chinese characters are very complex and diverse.

At the beginning of the computer, engineers and designers agreed to use a 5X7 bitmap grid to create low-resolution English digital fonts. In this way, the size of each character is about 5 bytes, and the computer's memory will not have too much burden.

All 128 low-resolution characters stored in the American Standard Code for Information Interchange (ASCII), including every letter in the English alphabet, numbers 0-9, and common punctuation marks, require a total of 640 bytes of memory. At that time, Apple's second-generation memory was 64KB, which could easily carry the English font library.

▲ASCII character table

However, due to the complicated glyphs in Chinese, it will be confused in a 5X7 grid, making it difficult to recognize. Therefore, at least a 16X16 or larger grid is required.

In this way, the size of each Chinese character is at least 32 bytes. If you pack 70,000 low-resolution Chinese characters, at least 2MB of memory is required. Taking a step back, even if only 8000 commonly used Chinese characters are put in the font library, about 256KB of memory is required.

This is undoubtedly a big problem. Because in the early 1980s, the total memory capacity of most PCs did not exceed 64KB, which simply couldn't fit the huge Chinese bitmap font library.

▲Photo of Sinotype III monitor, showing Chinese fonts. Picture from: Stanford University

The memory shortage is not the most troublesome, because it can be solved with the advancement of PC software and hardware. How to create a recognizable and beautiful Chinese font in a 16X16 low-resolution grid is a more difficult problem.

To this end, the designers of the Louis team spent several years trying to create Chinese bitmaps that meet low memory requirements, are clear and easy to recognize, and even have calligraphic beauty. Among them, Huan-Ming Ling (Huan-Ming Ling) and Ellen Di Giovanni (Ellen Di Giovanni) made the most outstanding contributions.

They first used paper, pen, and correction fluid to draw the bitmaps of Chinese characters, and then used the Gridmater program mentioned above to digitize them and implant them into the Sinotype III system.

▲Chinese font displayed by Sinotype III. Picture from: Courtesy of Bruce Rosenblum

The craftsmanship behind the production

In the archives, Professor Thomas discovered the whole process of the Louis team designing the bitmap of Chinese characters. In a booklet full of grid graphs, it records how designers create bitmaps of Chinese characters by hand-painted scatter symbols.

We all know that the strokes of Chinese characters are not "horizontal and vertical". The entrance strokes, exit strokes, and stroke gradients all have rich details. This is also the core problem faced by designers, that is, how to show the beauty of calligraphy as much as possible in the 16X16 grid.

In this grid book, you can find that every Chinese character has been carefully drawn by the designer. The green "X" is the original mark. After it is reviewed by the Chinese character editor, if something is not standardized enough, Louis and his team will cover the original mark with correction fluid, and then mark it with a red "X".

▲Draft bitmap of the word "Back". Picture from: Stanford University

After repeated modification, the final confirmed bitmap will be input into the system.

If you want to meet the needs of consumers, the font database must contain at least 3000 commonly used Chinese characters. This amount of engineering is huge for the team. People might guess whether they will look for some tricky methods.

For example, for Chinese characters with the same radicals, you can directly copy the radicals. Just like the words "review" and "reading" in the picture below, it stands to reason that designers only need to design the different parts on the right side.

▲Draft Chinese bitmap. Picture from: Stanford University

But Professor Thomas found that there were few similar working mechanisms in the archives. Louis insisted on requiring designers to adjust and design word by word to ensure that the radicals of each word look coordinated. Even if some changes are very subtle, it is hard to detect.

Professor Thomas reproduced the Chinese font of Sinotype III according to the file information. It can be found that the "juan" and "birth", which are also the same as "女", have different designs besides the word "女".

▲Can you see the difference next to the word "女"? Picture from: Stanford University

The width of the word "juan" beside the word "女" is 6 pixels (grid), while the width of the word "delivery" is only 5 pixels. In addition, the skew points and strokes of the word "女" in the word "Morning" are one pixel more than that of "Juan", which is visually longer.

Such meticulous design is not alone. Professor Thomas found a lot of similar work in the font library. When he compared the draft of the bitmap with the final product, he could see many subtle and interesting changes.

For example, in the character "Luo", the stroke in the lower left corner initially stretched downward at 45°. But in the final version, the end of the stroke is "flattened", which is more in line with the artistic sense of calligraphy.

▲Two versions of "Luo", the left is the final version. Picture from: Stanford University

It can be seen that adding or reducing a pixel will affect the overall sense of balance and beauty. This also reflects the hardship of the designers in creating the first batch of fonts and the craftsmanship behind them.

In fact, the 16X16 grid is not very friendly for creating Chinese fonts. The main problem is symmetry.

We know that a large number of Chinese characters are symmetrical, and according to the rules of mathematics, only an odd-sized space area can create a completely symmetrical shape.

Therefore, Louis and his team decided to use only the 15X15 area in the 16X16 grid to achieve symmetry of Chinese characters. This further reduces the designer's space and puts forward higher requirements for the design work.

▲ The symmetry and asymmetry of mountain, middle and field. Picture from: Stanford University

Thanks to the team's tireless efforts and meticulous attitude, Sinotype III's Chinese font library project was successfully completed. Although it has not been released commercially, it is indeed one of the first PCs in the world that can process, display, input and output Chinese.

Of course, the way Louis and his team make fonts seems too old-fashioned and naive in today's technical context. The widely used TrueType font technology can store fonts in a vector format, with a small footprint, fast rendering, and clear and sharp display effects.

▲Nowadays most fonts are in TrueType format. Picture from: themex

But it was they who used the "stupid method", verbatim drawings, and repeated revisions that allowed Chinese characters to enter the digital world. The high-resolution glyph information compression technology developed by Academician Wang Xuan of "Contemporary Bi Sheng" completely solved the dilemma of Chinese character encoding storage.

Thanks to the efforts of these predecessors, Chinese has not been let down by the Internet tide, and the theory of Chinese Latinization has been swept into the historical garbage dump. We are able to surf the Internet in Chinese today, and we should thank them for their youth for this.

TalkFresh. Work email:

#Welcome to follow Aifaner's official WeChat account: Aifaner (WeChat ID: ifanr), more exciting content will be provided to you as soon as possible.

Ai Faner | Original link · View comments · Sina Weibo