Font converter

WOFF2 is the same outlines as a TTF or OTF with substantially better compression built for exactly this purpose — typically thirty to fifty per cent smaller than the original. Every browser in current use supports it, so serving anything else on a web page is spending a visitor’s bandwidth for no benefit.

TTF and OTF remain what you install on a computer and what design software expects. Converting between those two is largely a repackaging: the difference is in how the glyph outlines are described, and both work everywhere a desktop font works.

Converting a font does not change what you are permitted to do with it. Many typefaces are licensed for desktop use only, and putting one on a website is a separate licence that the conversion does not grant. Open-source families are the exception and are usually distributed in web formats already.

This is worth a sentence because the conversion is trivial and the licensing is not, and a tool that quietly makes the first easy should at least mention the second.

TTF and OTF are the desktop formats — what you install on a machine so applications can use the font. OTF can carry more sophisticated typography, but for installing and using, the two are interchangeable in practice. WOFF and WOFF2 are the web formats: the same outlines wrapped in compression, meant to be downloaded by a browser rather than installed.

WOFF2 compresses roughly 30 per cent better than WOFF and is supported by every browser in use. There is no longer a reason to serve WOFF as a fallback, which was standard advice until about 2019 and is now simply extra bytes.

A full font covering Latin, Cyrillic and Greek with every symbol can be several hundred kilobytes. A subset containing only the characters an English page uses is frequently under thirty. That is a far larger saving than any format change, and it is the single most effective thing to do to a web font.

The risk is cutting too close. A subset built from one page will break on a page that uses a character it lacks — an em dash, a curly quote, an accented name in a comment — and the browser silently substitutes another font for that one glyph. Subsetting to a named character range rather than to observed usage avoids the trap.

Hinting is instructions inside the font telling a renderer how to snap outlines to a pixel grid at small sizes. On the low-resolution screens of the 2000s it was the difference between crisp and blurry text, and manual hinting was skilled work.

High-density displays and modern rendering have made most of it unnecessary, and conversion tools frequently discard or regenerate hints. On a retina-class screen nobody will see it; on an older Windows machine at 12 pixels it can still show. It is worth knowing the information can be lost in conversion, and worth not worrying about it unless the audience is on old hardware.

A variable font carries a continuous range along axes such as weight, width or slant, so one file provides everything from thin to black rather than a separate file per weight. For a page using four weights that is usually a substantial net saving.

Conversion between formats generally preserves the axes, and conversion to a static format does not — it produces a single instance at whatever coordinates were chosen. If a font file is unexpectedly large, being variable is often why, and it is a feature rather than a problem to compress away.