Chinese is at the bottom.
The problem I see with the assertion that the fonts are changing depending on the size they are displayed is how to actually encode it. Not that it’s impossible, but how would it be done?
For our friend, the character in question, it’s obvious that there is one code which is then rendered differently depending if the font set is Japanese or Chinese.
In Japanese, the 目does not go to having one internal line (three lines total) to two internal lines (four line total) no matter how small it gets. Nor does this happen on my Chinese iPhone.
If this is happening, how would it happen? I guess the program could select the rendition depending on the size, but why? If it’s small enough to require four lines to be changed to three, does it really matter if they are all displayed?
This is something which I never encountered in Japan. I was involved the documentation and printing back in the early 90s, when computer and display resolutions were much lower and wasn’t aware of it. Not that it means it didn’t happen.
On further thought, this would certainly only be related to bitmap fonts for computer display, as any printed material uses vector fonts.
For the original question, my (former) computer graphic artist was a wizard when it came to Illustrator. There were times when he had to tweak things, but it would always be done on the computer. I can’t really ever see a need for hand rendering and then scanning something like this.
Scalable fonts may have different representations for small sizes that deal with limited resolution. At a small enough size a designer might choose to eliminate a line to make a character fit. And when scaled from only a single representation lines could disappear from the scaling process. It can happen in any font, and it’s more likely in the complex and densely formed Chinese characters.
When roman character typesets have thousands of characters, that’s because they’re well made. They may include cyrillic in the set to make up a large portion of characters, but any good font will have ALL possible accent marks built in. That means a forward accent mark on a capital O, and a backward one, and an umlaut, and a tilde, now repeat for every other letter that has those…
Then, any good font will also have ligatures. When you type “ruffle”, a good font will combine the “ffl” into a more friendly format where the three letters run together (the program you’re typing in has to support it - for instance, the web does not). Then if it is an especially good font, you will get non-standard fancy ligatures, such as “st” where the t and s dramatically curl their tops together. A good font will often include superscript, subscript, and smallcaps as well (programs can “force” this by artificially changing the normal character size, but that is frowned on). It may include special fractions.
If it is in an italic or handwritten style, the best fonts today have hundreds of alternate characters that fill in as you type to make the font look more personal and handwritten so that not every “t” and “l” is “written” the same. You may be able to choose between 4 different capital A styles and words like “the” “of” and “and”, may have super-special characters worked in for the whole word as a block. When you type “party” it will see that and make the y fit in a special way that wouldn’t have happened if you typed “fancy”.
So, in reality, it’s not that different than the thousands of Chinese characters one has to make, if you’re making your fonts properly. I’m sure font designers are daunted by how much work goes into a Chinese font set, which is why there are fewer of them than roman sets. I wouldn’t be surprised if Chinese sets get less fancy upgrades due to their scale.
A good designer will make sure that a font reads well at any size - which often means specific weights for caption sizes and specific weights for billboard sizes, as they read totally different due to scale. A caption weight font will look very blobby when scaled up to 2 feet high. And if it’s specially made for print use, they sometimes put in special gaps where lines meet up (like the white space under the spikes of an M) because ink “traps” in those gaps and makes the letters look smudgy when printed. There is a LOT of work that goes into even roman fonts, and there is simply more work for Chinese fonts. As they wrote them all by hand before, so they do again, except now they pull it into a vector-based editor and fiddle with the lines for two years making it perfect. They usually write it by hand on actual paper initally for fonts that are supposed to look hand written. All font designers learn how to draw even clean serifs like Times New Roman by hand first as a part of their training, though they’re made all on the computer in practice. It’s one of those things like learning long division and then next week the teacher says “oh, and you can do this on the calculator from now on”.
This is why fonts taken from places like DaFont are so lackluster; if they’re not a specialty font (like something that looks grungy for Halloween or something) they often lack any of the requirements people need out of a professional font. Fonts from DaFont have mere 52 letters plus extras like ? and ! - if you’re LUCKY. If they’re a “normal” looking font, 9 times out of 10 they are poorly made.
Oh yes, and then repeat everything above for the semibold style, the semibold italic, the bold, the bold italic, the bold extended, the semibold extended, the thin, the thin italic, the caption size, the poster size…so on and so forth, depending on how versatile they want the font to be. Every width has to be specially altered. Simply throwing a fattening outside line on the letters is not good enough. Then you have to check that your umlauts aren’t touching the tops of the o’s and then kern the letters so that when someone types “1950” it doesn’t look like “1 950”…
This is why full professional font sets often run $35 for one font weight, and hundreds for the full set.