How do you create alternate font glyphs/characters?

Question

I used a font recently that had alternate glyphs assigned to each character. It was a handwritten font. Like someone scanned and traced a sample of their own handwriting. And I guess to make it look more convincing, each letter was scanned/traced several times.
So if you type the uppercase letter 'A' (or lowercase 'a', or whatever) four times in a row, it could appear as four similar, but slightly different (or sometimes quite different) variations. But it was only a single font. It also seemed to prefer a particular glyph variant at word boundaries. So, the first and last letter of each word might have elaborate flourishes or some fancy thing like that. In Adobe Illustrator, the alternates pop up as you type, and you get to pick & choose with the mouse.

Maybe this isn't uncommon, but it's the first time I've seen it. Anyway I'd like to create a new font of my own, or perhaps modify an existing one to utilize alternate glyphs in the same way. But how?

For example, this font is designed to look like the one from the iconic Coca Cola logo. But it can't really recreate the original because the original has two different "C" glyphs. Maybe that's not the best example, but you get the idea.

---

Answer 1

In general: you need to use OpenType features to make “intelligent” automatic replacements. So it needs to be an OpenType font. It can have a TTF or OTF suffix.

Taking your Coca-Cola example, there are two ways to achieve this effect:

1. you replace certain strings with ligatures. This can be two characters like Co or entire words (like Cola) or phrases (like Coca-Cola). Standard ligatures go in the LIGA feature, optional ones often go in the DLIG feature. The coding logic is: If f + i, replace both with fi ligature.

2. The other option is contextual variations. In this case, you don’t replace the entire search string, but just parts of it, based on the surrounding. For example: You could have a C without any swashes as default and then two variations with a swash to the right at the bottom and a swash to the right at the top. The replacement is then done only if there is enough room. If a y follows, you can’t use the bottom swash; if an h follows, you can’t use the upper swash. To achieve this, you need to create so-called classes of relevant characters, e.g. “has_ascender” and “has_descender”. Then you make the substitution in a feature like CALT or CLIG. The coding logic in this case is: If C + [has_descender], replace C with C with upper swash.