Variable OpenType test fonts
Voto Serif GX (OFL)
The Voto Serif GX is a variable OpenType font based on Noto Serif & Noto Serif Display. Noto Serif serve as the
opsz axis master =12, and Noto Serif Display as the
opsz axis master =72.
ttffolder has the working font in Variable OpenType TT (.ttf) format.
srcfolder has some sources and the
toolsfolder has little helper tools.
00-originalfolder has the original
.glyphssources, with 2 axes each (weight and width), having 4 masters each (same for Italics). Most glyphs in the uprights are compatible, some have more or less trivial compatibility issues that should be resolvable via prepolation. (Note that there are interesting cases e.g. in one font a glyph is composite, in another outline-based).
01-mergedfolder has the sources for the original designs, which have the
wdthaxis bounds 70 and 100. In the Voto Serif fonts, compared to the original Noto fonts, some incompatible glyphs in the uprights have been compatibilized, some have been removed. But the glyphset is still large, >3,000 glyphs.
02-wdth-extrapolfolder has the sources extrapolated to
wdth50 and 130, and the number per axis reduced to 3 instead of 4 (so 12 masters instead of 16). I’ve reduced the number of masters because of UI deficiencies in Glyphsapp, but also for reasons of practicality.
makeInstanceMapForGlyphsapp.pytool is included which produces a handy list of instances which can be pasted into Glyphsapp’s Font Info. With this tool, I’ve made 7 x 9 x 9 = 567 predefined instances that are included in the font (and in the
02-wdth-extrapolsources). The instances are named
<pointSize> <widthClass> <weightClass>e.g.
24 5 400, rather than using traditional text descriptors. This number of instances may be exaggerated but is relatively realistic for big font projects. There are in some design deficiencies, especially in the Light masters, due to the extrapolation, but these can be ignored.
I tried building the fonts with Google’s
fontmaketoolchain. Currently, only the upright font can be built, and even that not fully — without any
fontmakefails when trying to merge
kern, regardless of whether built using
afdko(I had to tweak
autoFeaturesin order to build the fonts). This font presents a good test case for authors of tools that do
GPOStable generation and blending for variable fonts.
Despite the deficiencies, thanks to the wide range of
opsz axis, and a large number of predefined instances, the font can serve as a good source of experiments. Note that the font does not yet contain the
STAT tables required by the OT 1.8 spec for variable fonts.
Copyright © 2016 by Adam Twardoch and the original authors.