Iteration to compatibility

googlefonts / roboto-classic

Development of a Roboto Variable font

SIL Open Font License 1.1

151 stars 15 forks source link

Iteration to compatibility #38

Open dberlow opened 4 years ago

dberlow commented 4 years ago

Hey Mark,

Mike Duggan has joined with Font Bureau as a subcontractor to work on compatible output between Roboto VF and the Roboto TTF instances, and optimization of the VF’s size in whatever ways are possible.

Mike has produced his first and second hinted candidate vf, and we’d like to begin iterating, using the test software and formats you use, to take that over ourselves until we iterate to compatible candidate.

We can then present a comparison of the hinted, unhinted and legacy fonts, work on optimization and produce a version with only a wght axis, but compatible with the legacy Roboto weights only.

So, Mike and I are on this issue, and when you have time, please lead us once to or through the tests we’ll need to pass.

Thanks, David

m4rc1e commented 4 years ago

I've just compared the font in this dir, against the original hinted v2.136 statics.

Remove MVAR table
fvar coordinates for wght axis are incorrect. All instance should have values which are multiples of 100.
hhea.ascent should be 1900, hhea.descent should be -500. We've got hhea.acent 1638 and hhea.descent -512.
We should use DaMa statmake to produce a proper STAT table.

I also expect the build chain of this project to be of a similar quality to the original Roboto repo. We can finalise this once we're happy with the fonts.

Regarding the hinting, I can't fault it atm. I'll update this thread once I've examined the results more thoroughly.

m4rc1e commented 4 years ago

Jumps only occur past 20pt. I think this is acceptable. There are other very small changes such as the counters of the "B" or crossbars of certain letters may be at a different heights. Overall, they are improvements.

Diffs are also very clean so I'm pretty happy. The overall quality is higher than Oswald which is our most popular VF so I'm pretty optimistic.

The only issue I've noticed is that the linespacing is different on Win Chrome 69 vs Win Chrome 79. I still do not know what's causing this. The os/2 version in the VF is v4 whilst in the old statics it is v3.

I suggest we fix the issues I found above and I'll investigate what may be causing this line height issue. cc @davelab6

davelab6 commented 4 years ago

@m4rc1e are all the above issues fixed? If yes please close

m4rc1e commented 4 years ago

I'm on it. Please be patient