SSA/ASS font subset tool (shrink your fonts)

[pie]

I made a small utility to take some SSA/ASS subtitles and fonts, parse it, and create stripped down fonts containing only the characters required (font subsets).

e.g If using only one character from a font, then it will strip the font so it too only contains that one character.

Link: dead

I found this quite useful for fonts that contained CJK characters (which are often ~3MB+ each) - practically all fonts are reduced to under 100KB each, and generally I found I could get most scripts working with fonts of a combined size of 150KB or less.

Basic usage:

1. Set mkvmerge/mkvextract paths:
typochop -p "mkvmerge:/path/to/mkvmerge" "mkvextract:/path/to/mkvextract"
2. Feed it input (both fonts and subtitles):
typochop -o . video.mkv
typochop -o subsetted-fonts font-folder subs.ass

Some interesting use cases:

1. Collect all fonts for a particular series into one folder.
2. Place all subtitles for that series into another folder
typochop -o subset fonts-folder subtitles-folder
This will generate font subsets specific to each episode.

1. Do as above, but use this for the command instead:
typochop -a -o subset fonts-folder subtitles-folder
This will generate one set of fonts that will work for any episode in that series.

This tool is written in Java and uses sfntly to work with the fonts.

[pie]

A new (beta) version. (link dead)

Some improvements from last time - (hopefully) more informative output, correct handling of drawing commands (ignore {\p1234...}{\p0} blocks), correct (I think) handling of hard spaces \h. It works pretty much most of the time on the stuff I've thrown at it.

Anyone ever use {\fe} commands, or have a sample that uses it?

[Plorkyeran]

Originally Posted by pie

Anyone ever use {\fe} commands, or have a sample that uses it?

Aegisub ignores it entirely and no one has ever complained. I don't think there's any valid reason to use it anymore.

[cyberbeing]

Use of either \fe or \fn is needed with VSFilter when a style font does not contain certain Unicode glyphs because of limitations in GDI fallback behavior.

Example: http://www.sendspace.com/file/w3662w