Some questions for Chinese type designers

Vasil Stanev

Hi,
1. I can imagine there are many hanzi that are only used in historical context and do not need to be included in non-specialized type outside of Akademia. Are they grouped in ranges, is there a list, and are you guys completely sure how they should look or what they mean? I imagine there are many hanzi that describe various tribes fighting the empire, obscure beliefs and practices, historical items, died-out ideas and all kinds of such stuff - it's a very rich history.

I am asking this because I still see people blindly going headlong and designing modern versions of Cyrillic letters like юсъ and ятъ that have been obsolete for a very long time and are not used anywhere - and when they are used, there are fonts just for that. Something similar is perhaps the case with Oracle bone script?

2. why is the one end of a horizontal stroke slightly raised above perfectly horizontal? are there similar obscure rules about the other main stroke types?

3. can you be a good CJK type designer without being a good calligrapher?

4. how do you keep kerning to not make the font too heavy. How many kern couples are there before it breaks down?

5. BESIDES simplified hanzi and other local differences like radicals, different elelments etc, are there minute differences between Japanese and Chinese renderings of the same character? Between the same hanzi and kanji? Something similar to what I describe in 2. ?

6. in what way do you guys use the PUA? I can imagine the biggest presence there are the hanzi since they have the biggest glyph pool to draw from.

7. is there a .locl feature in a CJK font?

8. do the Vietnamese and Koreans still use hanzi on rare and special occasions? Are they the very same?

9. Is 15 characters per day a reasonable quota for a single member of the team? What are the time-saving techniques besides radicals as components? Are there specialized scripts, software or the same to fasciliate work?

10. Are there many fully cursive fonts (don't know the term. I am trying to describe a style where the brush does not leave the silk before the character is finished, and it's written very fast)?

11. what is the name of the style when the characters looks like the thing it describes, i.e. the hanzi for heron looks like a heron.

12. please list the ten most used and best, in your opinion, hai fonts.

13. Do you feel there will be a boom in CJK fonts. And then maybe over-saturation and bust?

Thank you in advance! Feel free to point me to online resources, preferably in English, Russian or German. 

Aaron Bell

Not a Chinese type designer, but I might be able to give you some answers. 

2. You’re referring to characters designed based on calligraphic traditions. When using the brush, strokes tend to bend a bit rather than being perfectly straight.

3. I’d say you don’t need to be a good calligrapher, but having a good understanding of brush-structure and application is very useful, as there are brush-inspired aspects to design that one should pay attention to.

4. To my knowledge, CJK fonts do not use kerning. Characters are designed within the Em-square (or rectangle) with space on either side of them. At least in Korean, folks don’t seem to mind if certain syllables overlap a little.

5. Generally not to my knowledge. Most fonts that include different CJK writing systems share Chinese characters between them (except for localized variants).

6. PUA shouldn’t be used by anyone for anything . There are buckets of unicode slots for Chinese characters (CJK through CJK extension F). Why need PUA?

7. Sometimes. <locl> features can be useful for accessing differentiated characters between Japanese, Korean and Chinese (Trad & Simp), but due to the ever increasing file size, time investment, and costs of merging all of these together, it usually makes sense to focus on just one of them at a time—unless there’s a very compelling reason to do so. Adobe Pan-CJK font is one of the few that has it.

8. Not sure about Vietnam, but Korea does still use Hanja on occasion. In fact students are required to study 1800 of them. Like Japan, there are regional variants for Korea. But otherwise, they’re similar.

9. There are specialized scripts / software, but really it is just a lot of brute force. For larger foundries, there’s a division of labor between the designer, the digitizer, and production, so each focuses on one part of the design process.

That said, there’s some efforts for acceleration. Arphic Design in Taiwan has a stroke-based mechanism for drawing fonts, but it is internal and requires licensing their software to even use the font file. At ATypI in Poland, Morisawa presented that they’re working on AI to help accelerate the drawing process to help speed things up.

13. IMO, there can be a boon if the speed of production can increase. But there’s also the eternal question of getting people to pay for fonts. So, maybe? We’re still plugging along in Latin with endless new fonts right? 

Ray Larabie

1: The problem with subletting is knowing what to subset. There are some obvious historical ranges but as for the bulk of it, it';s hard to know what to skip. In Japanese, a lot of Kanji people don't know the meaning 

2: Obscure rules. Some rules aren't so obscure because they're based on a radical which, without that angle, would probably be misconstrued. If you examine a more techno looking typeface and the stroke had an angle or bend, it probably means it needs to be there. The radicals within a character have their own meaning but they're abstracted in different ways. For example, the character for heart looks much different on its own, when it's squeezed tall on the side, or squashed at the bottom. There's some leeway to make it aesthetically pleasing but you have to know the threshold where it might be misconstrued as something else. This one drives me crazy...why is it so different from how I'm taught to draw it? But then other characters with what looks (to me) like the same radical doesn't have the same abstraction?

13: I don't think there'll be a boom in CJK fonts soon because of how expensive it is. You can reduce some of the development time but it's still going to be very expensive and take years to develop. If it takes a team of five, four years and a million bucks now, reducing that by half still represents a major investment. Compare to the 1990's boom for Latin fonts where you only needed about $6000 (in today's money) in software and hardware, one person and a few months to create something commercially viable. As I understand it, the testing requirements are a lot higher for CJK fonts than they are for Latin. In Latin, you're probably not going to have to deal with situations where a letter is misconstrued for another letter...not late in the process anyway. But with each kanji radical, there's a threshold that's different for each reader. For example, the same structure with different stroke lengths can change the meaning. But how much do you need to differentiate the stroke lengths to make it understood? Is it a radical that people are going to understand from context anyway? What about when it's on its own? 85,000 potential problems that need extensive testing and revision. For shared Chinese/Japanese characters, you need to know how the rules can be bent.

Vasil Stanev

Thank you.

Ray Larabie

Sorry...I didn't finish this sentence.

In Japanese, there a lot of Kanji people don't know the meaning of but are used in place names and people's names. For example, my local subway station, Aratamabashi 新瑞橋 has a character (second one) that out-of-towners often don't know how to pronounce and don't know the meaning. Even if it's an obscure character that's only used in names and very rarely, what if someone with this name becomes a celebrity and that name is in everyday use? So subsetting is very hard to do...there's always someone using a weird kanji for a name or place.

David W. Goodrich

You might want to look at "Source Han Serif/Noto Serif CJK History & Development," and related entries on Ken Kunde's CJK Type Blog.

David

Hin-Tak Leung

Speaking as a native language user,

1. You can follow the older big5 (for Taiwan), gb2312 (mainland) standards for commonly used characters. The HK government published a HKSCS additions for characters outside big5 which are in common use. The Japanese government divides the characters set as JIS level 1 and level 2.

2. That's mainly historical - left-handers had never been recognized as an accepted variant in the society so everybody writes with his/her right hand. There are some advantages of not getting your sleeve dirty writing this way, for example.

5. Yes. Glyph shapes diverged over the course of history. 

6. Before unicode, PUA was used for many of the rarer local variants.

10. Calligraphy was historically regarded as an art firm on its own, so there are different schools/styles.

About 2 and 10: the way characters are formed is influenced by the tool / material available - kai and the cursive styles are based on hand-writing, so have the tilting of the horizontal strokes. Ming/Sung is based on wood-cutting on movable types, so they are properly horizontal/vertical, with endings which avoids damage/splitting of wood. Oracle scripts were cut on bones, seal scripts were stone crafting, etc.

Hin-Tak Leung

Oh, the Taiwan government also have a guideline document on 5, about characters which are often written the wrong way (I.e. In HK/mainland).

Vasil Stanev

Thank you!