Plan for Noto CJK / Source Han Kaiti (楷体) and Fangsong (仿宋)

[stone-zeng]

For CJK script (especially Chinese), Songti (Serif, 宋体, 明朝体), Heiti (Sans, 黑体, ゴシック/), Kaiti (Regular script, 楷体) and Fangsong (Imitation Song, 仿宋) are four basic kinds of typefaces, see clreq. Four frequently-used Typefaces for Chinese Characters.

In the Noto super-family, there are three versions of Arabic fonts: Kufi, Naskh and Sans. Similarly, except for the current available serif and sans version, is there any plan for Kaiti and Fangsong for CJK script? Or maybe in Adobe's Source Han family?

[lhy7889678]

I agree to such a plan. Up to now, it seems that we do not have any higher-quality FOSS Kaiti and Fangsong fonts. Actually, we do have Fandol and cwTeX fonts, but there are some issues which may not make them really safe "free" fonts, and they lack language variations, multiple weights, and a rich coverage.

However, I understand this may be tons of efforts, and may not be achieved in a short time. In my opinion, Kaiti and Fangsong fonts are more difficult to design than the existing Songti (Serif) and Heiti (Sans). Unfortunately, I am busy with my schoolwork now so I may not be able to contribute to the plan much. But we can wait!

If we do have such a plan, I think we should make it as high-quality as the existing Noto CJK / Source Han fonts. Here are my suggestions, which may be or, well, not be really useful.

1. As for the non-CJK parts, we can try the existing EB Garamond for Kaiti and Roboto Slab for Fangsong. (Note that while EB Garamond is OFL-licensed, Roboto Slab is Apache.) They have variable fonts, which means we can fine-tune them to match the CJK parts.

2. If it is really hard to design, we can try teasing the Katakana part of the Songti font into Kaiti glyphs (which may be a bad idea :smile:).

[oldherl]

We do have FOSS Kaiti fonts: AR PL KaitiM GB and AR PL KaitiM Big5, and other derivatives such as UKai. Those were donated by Arphic company long ago, and are now included in many Linux distros.

[lhy7889678]

@oldherl Oh yes, they are. Nevertheless, I haven't found a FOSS Fangsong font (other than Fandol) yet...

[ghost]

I am also in need of a high quality Fangsong font for Japanese. In Japan, Fangsong is known as "宋朝体"(Sōchō-tai) .

[nengxu]

For CJK script (especially Chinese), Songti (Serif, 宋体, 明朝体), Heiti (Sans, 黑体, ゴシック/), Kaiti (Regular script, 楷体) and Fangsong (Imitation Song, 仿宋) are four basic kinds of typefaces, see clreq. Four frequently-used Typefaces for Chinese Characters.

In the Noto super-family, there are three versions of Arabic fonts: Kufi, Naskh and Sans. Similarly, except for the current available serif and sans version, is there any plan for Kaiti and Fangsong for CJK script? Or maybe in Adobe's Source Han family?

Support this plan

[xsrvmy]

Actually, there is the Klee One font for Japanese which is somewhat a mix of Kaiti and Fangsong. There is also a Mainland Chinese (SC) derivative called LXGW and a (partial) Taiwanese (TC) derivative called ian-sui both on github. They don't fully follow regional standards but do fix the most obvious issues between Japanese and Chinese.

[Tosakun]

Since all those fonts are open source, wouldn't it be possible to contact the font creators to use them as a base for Noto Regular/Source Han Regular fonts? Most of the necessary glyphs are already there as Klee covers both Japanese and Korean through opentype on Adobe Programs, while LXGW and Iansui cover Simplified and Traditional.

The only real sticking points I can find (aside from Time, Resources, and other things I don't have knowledge on), are that in LXGW 亠 and other characters like it like 宀, the dot on top follows the Japanese style not Chinese. Also some characters like 糸 have serif like protrusions in LXGW and Klee, unlike the more the more handwritten equivalent of Iansui. Examples below. (Klee, LXGW, Iansui)

Another thing is that the Hong Kong variant would have to be rebuilt as I don't think there is a Hong Kong font that derives from Klee.

Proof of concept:

[Tosakun]

Should mention that I'm interested in this as someone who is making stroke order diagrams, and having one consistent font for all regions would be so helpful for my OCD. Plus the only brush like fonts I found for Korean are Gungseo, which isn't very friendly for stroke diagram production, or don't use Korean Standard styles.

Plus it would help this little community https://commons.wikimedia.org/wiki/Commons:Stroke_Order_Project and their endeavors as they have an issue with finding free fonts.

[xsrvmy]

Regarding the LXGW font, that font treats certain stroke difference as design decisions, and does not fix them, where as Iansui matches the Taiwan MOE. The MS fonts do not have these design features

[walkthetalk]

Support this plan

+1

[KrasnayaPloshchad]

We do have FOSS Kaiti fonts: AR PL KaitiM GB and AR PL KaitiM Big5, and other derivatives such as UKai. Those were donated by Arphic company long ago, and are now included in many Linux distros.

However UKai lacks localized glyphs (locl feature). This could be fixed by making use of original AR PL KaitiM Big5 and AR PL KaitiM GB fonts, which still have glyphs to be borrowed, modified, and/or reintegrated to UKai font. On Debian site these fonts are still available from following packages: https://packages.debian.org/bookworm/fonts-arphic-bkai00mp https://packages.debian.org/bookworm/fonts-arphic-gkai00mp (On Microsoft Windows you can extract font file via 7-Zip.)

[Marcus98T]

Recently, there is a Noto Khitan Small Script in the Fangsong style, and I checked the Glyph sources, great news for font designers, they are overlapped sources. However, there are only nine proper Chinese characters in this font (again, only checking the Glyph sources).

But I'm sure that because the ancient Khitan small script borrowed Chinese characters for their phonetic letters (they composed them a bit like Korean Hangul from what I read) and did not modify some of them, those unmodified letters, which are identifiable Chinese characters, can be taken as they are and mapped to normal CJK Unified Ideographs, as marked in red.

Examples: 一 (U+4E00), 丁 (U+4E01), 干 (U+5E72), 王 (U+738B), 平 (U+5E73)

I wonder if a full-fledged open-source Fangsong font can be made out of this. Maybe even a multi-locale version, for example, a Japanese version as a Sōchōtai (宋朝体) font, and also an old traditional orthography version for Traditional Chinese, derived from the Japanese version, or forms closer to the Taiwan/Hong Kong educational forms, who knows?

Yeah, I know it will take a very long time if there are plans for this, so perhaps start with the Simplified Chinese version, and then work from there.

[KrasnayaPloshchad]

However, there are only nine proper Chinese characters in this font (again, only checking the Glyph sources).

They looks like Easter eggs. But I see it’s possible to create such Fangsong style font under free license.

[walkthetalk]

dear, 3 years past, any progress?

[Marcus98T]

Well, just a few days ago, an open source Fangsong font has been released: Zhuque Fangsong (朱雀仿宋). It's still in preview though, so no source is provided as of this writing.

On first glance, the font shape is absolutely not based on Noto Khitan Small Script, so I think Zhuque Fangsong has been worked on for a long time, even before the release of Noto Khitan and its source.

Soon the community may be able to expand it to cover more Chinese characters, with different glyph shapes for different regions.

With that, I guess Google won't ever work on Kaiti and Fangsong, obviously because it's too much work for them, they have other bigger priorities and someone else will likely do it instead. I'm not sure if this issue can be closed or not.

---

On the other hand, there is a SIL-licensed (or a licence equivalent to CC-BY 4.0, previously that was the only licence) version of Taiwan's Kaiti, 全字庫. I hope the community can eventually work on a version catering to Mainland Chinese standards and so on.

Quoted from the website (in Chinese):

1.本資料集授權使用者可依使用目的需求選擇『政府資料開放授權條款-第一版』或『開源字型授權1.1版(OFL 1.1) 』單一授權使用(非以上二種授權同時兼具)。有關授權說明網址：https://data.gov.tw/license。

Using Google Translate (and edited by me):

1. Users authorized by this data set can choose either the "Open Government Data License, version 1.0" or "SIL Open Font License 1.1" according to the usage purpose (but not both licenses at the same time). Website for authorization instructions: https://data.gov.tw/license.

[alicejjliu]

Seconding desire for a Kai and FangSong font for this series. It's really annoying looking around and not having a good FOSS source (except arphic, except arphic doesn't have a fangsong font i don't think), and many of which are missing parts of the unicode CJK plane. Even the article on google fonts talking about font classification in chinese doesn't have a kai or fang font that's FOSS.

[CarolinaFreitas]

Looking for a kaiti as well

[KrasnayaPloshchad]

Another option is get glyphs from this package and make derivations from that.