3 * Much less memory consumption while handling fonts with complicated glyphs.
5 * Option `-s' was partially broken.
10 * OpenType feature support. ttfautohint now uses the HarfBuzz library to
11 analyze data from the `GSUB' table. This allows the hinting of glyphs
12 that don't have an entry in a font's `cmap' table, for example
13 superscripts or small caps.
15 Related to this, the new option `--default-script' controls the default
16 (fallback) script used for OpenType features.
18 * More than a single standard character is used. For example, the `latin'
19 script uses characters `o', `O', and digit `0'. This improves the hinting
20 of fonts (and features) that have only a partial coverage of a script's
23 * Much better GDI ClearType hinting in the range 30-80ppem (approx.),
24 avoiding overly flat tops and bottoms of round glyphs.
26 * Better handling of non-square pixels (this is, horizontal and vertical
27 resolutions differ) in the created TrueType bytecode.
32 * Improved script support. Besides Cyrillic and Greek, which are now
33 handled separately from Latin, ttfautohint can handle Hebrew.
35 * Option `-f' now takes a parameter to specify the fallback script. The
36 corresponding long option name has been renamed from `--latin-fallback' to
39 * Work around a bug in display environments that use FreeType 2.5.0 and
40 earlier for rendering: Sometimes, the `strong' stem width routine was used
41 for DW ClearType (this is, subpixel hinting in FreeType is enabled) even
42 if `smooth' was selected while generating the font with ttfautohint.
47 * Option `--components' has been replaced with `--composites': By default,
48 the components of a composite glyph are now hinted separately, since tests
49 has shown that this gives good results in most cases. If this option is
50 set, however, the composite glyph itself gets hinted (and the hints of the
51 components are ignored).
53 An unfortunate side effect is that ttfautohint's option `-c' (which stays
54 as a shorthand for `--composites') now does exactly the opposite as in
57 * Older versions of Monotype's `iType' bytecode interpreter have a serious
58 bug: The DIV instruction rounds the result, while the correct operation is
59 truncation. This caused `exploding characters' with fonts hinted by
60 ttfautohint. Since many printers contain this rasterizer without any
61 possibility to update to a non-buggy version, ttfautohint now contains
62 work-arounds to circumvent the problem.
64 * Better support for glyphs where some points have almost the same position
65 (for example glyph `Oslash' in font `Roboto-Thin').
67 * Better support for glyphs which use explicit `on' points around round
73 * New option `--dehint' to strip off all hints without generating new hints.
74 This option is intended for testing purposes.
76 * Minor fixes to the created bytecode for compatibility.
78 * Minor GUI improvements.
83 * New option `--windows-compatibility' which adds two artificial blue zones
84 at vertical positions given by `usWinAscent' and `usWinDescent'. This
85 helps ttfautohint's hinting algorithm reduce the possibility of clipping
86 if those two values are very tight.
88 * Implement option `--x-height-snapping-exceptions', making ttfautohint
89 avoid x-height snapping for selected PPEM values. Useful in combination
90 with `--windows-compatibility'.
92 * Minor fixes to the created bytecode for compatibility and robustness.
97 * New option `--components' to treat components of composite glyphs
98 separately. This greatly reduces the bytecode size.
100 I'm waiting for reports whether this option works for most fonts; in case
101 this is true I'm inverting the option, making it the default (and the old
104 * Full support of TTCs, this is, all subfonts get auto-hinted now.
106 * The upper limit of the `--increase-x-height' option has been removed.
108 * Drag-and-drop support in the GUI.
110 * The TTY version of ttfautohint now acts like a (Unix) filter, this is, it
111 accepts stdin and stdout as input and output, respectively.
113 * Less memory consumption.
118 * A serious bug in the created bytecode has been fixed, causing incorrect
124 * A new, `strong' routine to handle stem widths and positions has been
125 added, to be selected with the `--strong-stem-width' command line option.
126 If it is active, stem widths and positions are snapped to the grid as much
127 as possible. This algorithm is useful for GDI ClearType support.
129 * A new command line option `--debug' (not available for ttfautohintGUI) to
130 print very detailed debugging information.
135 * The created bytecode has been reduced in size, making it approx. 20%
138 * New option `--symbol' to use standard stem height and width values instead
139 of using character `o' (which may be missing). Use this option for symbol
140 fonts or math glyphs.
142 * More documentation (in text, HTML, and PDF format). It's still
145 * Option `--ignore-permissions' has been renamed to `--ignore-restrictions'.
146 The short form is still `-i'.
148 * Defaults for various parameters have been set to more sensible values:
150 hinting-range-max: 50 (was 1000)
151 hinting-limit: 200 (was 1000)
153 * Option `--increase-x-height' now has a mandatory argument (in the range
154 6-20 or value 0 to disable it, default value is 14).
159 * Implement option `-x' to increase the x height of the font for small PPEM
160 values by rounding up far more often then rounding down.
162 * Add option `-G N' to switch off hinting completely above value N.
164 * ttfautohint now appends version information and the used parameters to the
165 `Version' field(s) in the `name' table. This can be suppressed with
171 * A GUI has been added, using the Qt framework. The binary is called
177 * The improved handling of composite glyphs in 0.6 was buggy under certain
178 circumstances, making ttfautohint crash and FontValidator complain.
180 * Dropout handling has been activated.
185 * Improved handling of composite glyphs.
187 * Implement option `-p' to pre-hint glyphs with original hints before
188 conversion takes place.
190 * Don't add a DSIG table if there is none in the input font.
192 * Human-readable error messages instead of hexadecimal error codes.
194 * Better tests (both at runtime and compile time) to reject too old FreeType
200 * Rendering on iOS is now expected to give good results.
202 * No bad rendering at very large PPEM values.
207 * The bytecode has been changed to `create' twilight points. This should
208 avoid rendering artifacts on some platforms.
213 * Fix font generation; sometimes the `glyf' table was one byte too short,
214 making the font invalid.
219 * Fix bytecode bugs which prevented correct rendering on some platforms.