Area JP: making of Japanese

Introduction

This article is a translation of the ongoing discussion between Caio Kondo (Inari Type) and Matthieu Salvaggio during the development of Area JP. Most of the records are based on the documentation Caio Kondo gave during the design process. The discussion happened on Discord and takes the shape of a back and forth between the two designers. Caio Kondo working on the development, art direction, production. Matthieu Salvaggio giving hints and questioning some steps of the process as he learns more about the way Kana are being designed alongside Japanese lessons he's been taking to understand the writing and speaking system better.

Early stages

Before I start drawing, I’ll begin by researching references. At this stage, it’s important to understand which Japanese style works best to bridge with the Latin script. Just like in the Latin alphabet, there are two main categories: Mincho (serif) and Gothic (sans serif). From these two, many other styles branch out. These references will be essential for defining proportions and metric alignment with Latin, which is why choosing the right ones is so important. Once we decide on the style to follow, the next step will be drawing the skeleton. From there, I’ll explain each stage in detail.

I’m sharing some material that connects Blaze Type’s fonts with different Japanese styles. This is only the first stage of the research, but I’ve already considered all of Blaze’s sans fonts, since I believe it’s important to bring diversity of styles into the catalog instead of focusing on just one. In the analysis, I organized the fonts from less humanist to more humanist, taking into account curvature and more fluid terminals. Just like in Latin, gothic kana also shows this subtle differentiation.

Area aligns with a squared style but shows some humanist details, especially in the terminals and a slight calligraphic presence, while still keeping a universal design.

The metrics & skeleton

At the same time, I’ll begin sketching the skeleton of a few characters. I like to start with frequently used grammatical particles such as の か は へ で と に, since they appear often and play an important role in shaping the structure of the system. Alongside these, the initial vowels あいうえお アイウエオ are also essential, as they help define the DNA of the project. Just like in the Latin alphabet, the construction process for kana is modular (though not a simple copy and paste haha). Certain characters act as key forms that feed into the overall design vocabulary of the system. This will become even clearer once I start grouping characters by their structural similarities. In this stage, I’ll also refine the vertical and horizontal metrics to harmonize both scripts. In my view, this is the most important part of the process!

This configuration allows uppercase letters to align with kana, so when the Latin script is rotated 90 degrees, the kana remains aligned, just like in these images:

At this stage, I’ve disabled all custom parameters for line spacing such as the metrics for ascenders and descenders.. That way, both vertical and horizontal writing keep identical line spacing. Later on, I’ll adjust this to avoid confusing text readability. Regarding the drawing itself, I’ve only sketched the skeletons of the letters without adding contrast or refinement. it’s still very early and will evolve a lot. The principle behind designing Japanese letters is the same across all Asian scripts: unlike the Latin alphabet, letters are visually centered within the box. It’s almost as if the letters were pieces of meat on a skewer. I try to keep the grid as clean as possible, and in the end, I check alignment visually using H H to see whether everything feels centered. This really helps reduce distractions and shows if things are visually balanced or not.

I’m still experimenting with proportions, especially trying to understand the best sidebearings for kana in order to match as closely as possible the spatial perception that Area provides.

On top of that, I also looked into circled numbers and letter characters. It’s important to include them in the unicode character as well. If they are encoded, Japanese users can easily access them through the IME input system for instance, by typing Maru (circle) + number, which then gets replaced with the circled number. Still, I don’t think we should exclude stylistic sets, since these features are becoming increasingly popular for non-Japanese speakers. They would make access easier for this audience too, rather than forcing them to copy and paste from a unicode chart or glyph map.

I’m going to keep the current stylistic set and also add the circled numbers and letters with their respective Unicode mappings. This won’t create any new stylistic sets, but these characters will then be recognized in software. That way, the software won’t replace Area with a default font when trying to access these characters. Here’s an image of that, in blue are the stylistic sets and in yellow my addition.

Hiragana Ki

MS: Do you think we should consider an alternate version of kiき where the under circular shape is joined? I know it can be separated as you did but the joined version exists as well, just pondering about it

CK: I hadn’t really thought about it, but I think that’s possible for this style! Which one do you prefer me to set as the default?

MS: Honestly I've seen both of the style, i suppose it's a good way to distinguish it even more from "sa" to have it unjoined?

CK: I don’t think I’d go that way; both should be perfectly legible. The unjoined version tends to show off the calligraphic style more than the joined one. Since this style sits between the humanist and universal styles, both can work! Personally, I prefer the unjoined version because I think it creates more interesting shapes. I like that it has alternates, since someone might feel the exact opposite of me haha

MS: I was wondering if the unjoined version was a link to calligraphic use or not. From what I've seen it is usually common in calligraphic / humanist kind of style sans. I think the unjoioned version would make sense to Area then.

MS: Have you tried letterinkapp for these sort of things? that could bring interesting results when experimenting with the strokes I suppose (not 100% sure its useful in our case with Area but for other style of fonts perhaps): https://www.lttrink.com/

CK: Nice! I think it could be interesting to use this for JP. I believe it’s possible to set up an ideal contrast for it. Thanks for sharing!

A few days later

CK: I’ve really enjoyed using the plugin to shape the letters. Since kana has such a strong calligraphic foundation, it’s been incredibly helpful. Thanks a lot for sharing it, it’s really speeding up my workflow. Right now, I’m defining the flow in the curves, as well as the stroke thickness and terminal angles. I’ll record a video soon to show how I’m doing it. In general, hiragana tends to feel visually heavier than katakana, which is why its strokes are slightly thinner. Balancing this is part of the challenge It’s still not refined, but I’ll start the refinement process once I finish defining this stage for all the letters. I expect to begin that later this week.

Discussion about tenten and maru (Japanese diacritics)

MS: Great ! Out of curiosity : is the い designed? Also I was thinking about the が (not « ga » but the diacritics changing the glyph prononciation): do you create them as component or manually do them for each Kana depending on their respective design? I am also thinking (let me know your thoughts about it): I think we should make those diacritics smaller so it is easier to distinguish the glyph change in prononciation. か が see what i mean ? I wonder if we should include some basic list of most common used kanjis too. I fear that without a basic set of Kanjis we might be lacking uses for Japanese speakers.

CK: Yes, the skeleton of い has already been drawn, but it’s not finalized yet. I still plan to refine the curves and adjust the terminals. The tenten and maru (Japanese diacritics) work well as components. Adjustments are only needed for a few characters that receive these accents, such as が, where I make small tweaks to make sure everything fits nicely within the box. I haven’t started working directly on the diacritics yet, however I agree that making them smaller could be beneficial, it helps with visual distinction and can also reduce the amount of adjustment needed for each letter. I’ll keep you updated once I move on to that stage. I understand your concern about the lack of kanji, but kana only fonts are actually quite common in the Japanese market. Pairing different fonts is part of the everyday workflow for Japanese graphic designers. Since kanji take a lot of time, people, and resources to draw, it has become very common for independent foundries to release kana only fonts. As I mentioned before, one of the most important phases is the metrics adjustment, since Area JP needs to fit well alongside other Japanese fonts that include kanji. This pairing fonts works very simply: users can set a kanji font as the default in their system, and all missing characters in the Area JP, in this case all kanji, will automatically be replaced by that default font. The list of kanji you mentioned make sense for people who are learning nihongo, since beginners might not fully understand how this pairing mechanism works. But for native Japanese users, this could actually make things more complicated, as the kanji in Area might look too different from those in the paired font. That’s why I think it might not be worth pursuing that direction, unless the goal is to target that specific beginner niche. Does that make sense to you?

Developing the system

MS: Completely understand what you mean. I didn't know the uses where as you stated for Kanji replacement. So, are numbers from 0 to 9 also something that is usually only kanji associated? « い » はうつくしいです。

CK: In our case, I did not include the Kanji numbers (二, 三, etc.) in the character map. This means that if users input these characters, the system will automatically use the fallback font. For other number forms, like circled numbers or full-width numerals, I added the corresponding Unicode values so that they display correctly without fallback

CK: The skeletons are now fully drawn, and to illustrate that, I recorded a short video of す, one of my favorite characters. It shows how I applied contrast by following the calligraphic flow and structure of the strokes.Drawing with this construction flow really helps me understand where more or less pressure should be applied to create contrast. That’s why having a solid grasp of calligraphic principles is so important in this process. Of course, I also rely on good visual references to guide my decisions.

CK: In your Nihongo (Japanese language) classes, you’ve probably come across images like this, that’s essentially how I’ve been approaching the entire skeleton process.

CK: Also I started refining the katakana. I always like to begin with the simpler part before moving on to the more complex one. Kata tends to be simpler because of its more geometric shapes, but don’t be fooled by that. Unlike Latin, there aren’t the same thickness values for all letters, everything depends on the amount of elements inside each box. The real challenge is to balance everything visually. Once I finish the hira, I’ll start testing text paragraphs to identify lighter or heavier areas. Now, you’ll understand why I mentioned before that the Kana construction process is also modular. Here’s as I usually guide myself during the design phase.

CK: Another important point, even though kata has less personality than hira, I’ve been thinking about ways to bring more Area identity into the J-face. I really like how the terminals are both curved and sharp at the same time, so I added these traits intentionally. What do you think about this direction?

CK: I also experimented with a slightly bolder approach for some characters. It does make them feel a bit too calligraphic, and it might not stay in the final version, but seeing it made me imagine how cool an ink-trapped version of Area could look!

CK: Looking ahead, I think we could explore this direction further with the Hiragana. I’ll keep balancing boldness with care when shaping these curved terminals, since not every character reacts as well as え, for example. Overall, I’m feeling much more excited now that things are becoming more refined! The beginning of the process is always a bit slow, but now it’s finally getting fun haha. Can’t wait to hear your thoughts!

MS: Yes indeed ! The ductus is 100% needed here (and most often cases for Latin glyph design too anyway). I understand the logic since Area has some mini curves in stems as you shown, maybe less visible and more discrete for Kana? I really like the idea you're inputting tbh !

The more I look the more I like the idea you're conveying. I think its going to be bringing a lot of corresponding thoughts in the Romanji/Kana approach. Really excited to see both languages mixed in text. Are you considering the Vertical design at this stage or is it probably better looked at later?

CK: I’ve made solid progress on the refinement process. It’s not fully finished yet, but the design is already looking much more polished. In the next stage, I’ll keep refining while testing it in text blocks. In these paragraphs, I’ve been exploring how the Kana interacts with the Latin, this has given me valuable insights for adding more personality to the design. I’ll also run some tests in vertical writing

CK: Here’s the latest update. I’ve been gradually bringing more of Area’s vibe into the kana, and here’s an example showing how that’s been applied. Also I increased the roundness of the kana shapes by about 20% compared to the previous version, I think this makes them feel more consistent with the softer, rounded qualities of the latin

CK: And I started positioning the accents. It’s still at a very early stage, but I’m not quite happy yet with the size or design of the tenten, so please don’t focus too much on that for now. Overall, the relationship between the kana and the Latin is working really well. Aside from the accents, everything is already reaching its final refinement phase. What do you think so far? I believe we can start moving into the weight expansion phase, if you agree.

MS: It really feels like the same font family ! I suppose the weight difference between Roman / Kana should be corrected once we start moving to bigger master and interpolate accordingly?

MS: Don't you think the second stroke of a is a bit short ? How about the third stroke, maybe going a bit too much on the left compared to the first stroke? I suppose you're going to work on those light part for these as you mentioned you wanted to work on the heaviest weight a bit more.

CK: Thanks for pointing these out! I agree, the third stroke is leaning a bit too much to the left. The second stroke is usually shorter in heavier weights, but I can give it a bit more depth if you prefer I think わ is consistent, but れ might be standing out a bit, since its upper curve seems a little higher than usual

CK: I made great progress this week, I feel like the weights are pretty much finalized. First of all, I wanted to thank you for the comments in your last feedback, they helped me a lot when refining those letters. recorded a screen showing how I approached the weight expansion. It’s nothing too mysterious: first I aligned コ with H to establish a consistent reference for both vertical and horizontal thicknesses. Then I used the hairline skeletons and expanded them with the new thickness

CK: After doing that for all the hira and kata, I went through a verification phase, looking for possible weight or drawing inconsistencies. I spent the whole week reviewing these details, these tests went back and forth until we reached the result we have now.

CK: After the refinement stage, I moved on to the accents. I brought in other fonts as references to confirm their correct placement. Some characters can be a bit tricky, so it’s easy to get lost, like ド/じ for example, where the accent needs to respect a certain distance. I’ve made the mistake of placing them too close before .In some cases, it’s necessary to adjust the letter itself so the accents fit well, in others, you only need to shift the component slightly to make room.

CK: Finally, I increased the thickness of the small figures by about 10%–30%, depending on the case, to make the text color more consistent, since these characters are used quite often

MS: Thanks for sending all the explanation along the way. I really love the な !

I'm wondering if the stroke of the " needs more weight. I feel like they are standing out a little bit and I'm wondering if they shouldn't be more related to the Kana in terms of optical consistency, what do you think?

CK: This week was all about testing and making the final adjustments. let me know what you think! Line spacing is set to work smoothly in both horizontal and vertical writing without causing any confusion.

CK: The accents are now optically more consistent with the letters, and you were right it really does look much better!! All OpenType features (full, vert, vrt2, etc.) have been reviewed to avoid fallback or issues. Really excited about the result, and very grateful for the opportunity and your patience. Let me know if we’re ready for the next typeface!

---