Yatra, Devanagari & Latin display

[Catherine Leigh Schmidt]

Hi everybody,

Yatra is a Devanagari and Latin display font based on the hand-painted
signage of the Mumbai local rail system. I started the Devanagari while
studying in India earlier this year, and the Latin is only getting started
now.

This high-contrast display face features angular cuts and open knots, and
mimics the heavy weight of much of the city's hand-painted signage. While
calligraphic in look, it features some choice idiosyncrasies particular to
sign painting. With the Latin I wanted to take a stab at a flipped pen
angle; I'm still working this all out. Suggestions and feedback more than
welcome!

You can find it here: https://github.com/cathschmidt/yatra-one

A technical note: This is my first time having to generate deva + latin
.otfs. I'm running ITF's python scripts (from
https://github.com/itfoundry/base-devanagari-gf) to do that but I don't
completely understand what's going on. For instance, it always wants two
.ufos in the /masters directory, and won't run with just one even if I run
the script with --nointerpolate. The two .ufos in /masters are the same
file with different suffixes. I'm also having a hard time figuring out how
to "turn off" all the generated weights (again, the interpolation) — you
can see the /styles directory is populated even though, in fact, there is
only one style. I've commented out everything except Regular in config.py
but it hasn't done the trick.

If anybody has any experience using this and could shed some light on
what's going on, please let me know. Despite this rather hacky approach I
haven't noticed any weirdness in the generated .otf (yet)



To do:
Finish the Latin lower & upper case
Match weights and proportions
Finish remaining Deva full forms
Spacing Deva full forms
Finish Deva half forms
Finish Deva marks & matras
Latin diacritics
Finish Deva numbers
Latin numbers
Special Deva conjuncts

Thanks,

Catherine

[Dave Crossland]

Hi!

Thanks for posting :) 

Regarding the OT development, I'm going to spend the first 2 weeks of December at ITF in Ahmedabad and I hope to work with Liang and team on better documentation for their build systems. 

I took a quick look at the image, and for the latin I have some suggestions:

- the 'c' is falling forwards, and is too similar to 'e,' I suggest rethinking it

- the 'f' neck join of neck to stem is jumpy, whereas in t and j its smoother, so I suggest refining it

- the 'n' seems slightly too narrow to me

- the 'o' and 'O' seem too narrow, and lacking overshoot  

- the top left corner of P and R seems too rigid; perhaps the flat top is good if there was a little overhang over the stem left edge? 

--
You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/b2246343-407d-48a6-9571-fc8ded45f623%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Catherine Leigh Schmidt]

Thanks for the feedback! Should have a revised Latin up today. 

Would loooooove some documentation for the ITF build systems — I would imagine I'm not the only one! ;)

To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-discuss+unsub...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/b2246343-407d-48a6-9571-fc8ded45f623%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Dave Crossland]

On 6 November 2015 at 23:54, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

Would loooooove some documentation for the ITF build systems — I would imagine I'm not the only one! ;)

Vesper Libre's glyphs file could also be helpful, Rob and Kimya put a lot of effort into making it possible to ship Latin + Deva fonts from Glyphs, and Vesper Libre's shaping support goes beyond what Adobe Devanagari can do correctly. 

Eczar is also worth a look, it is also FDK based and I believe doesn't have the python complexity of the ITF system. 

[Catherine Leigh Schmidt]

Today:

- lowered x-height for existing lower-case Latin glyphs. hoping it will make them more proportional with the deva.

- played with new shapes for c and e

- widened o

- new, more calligraphic shapes for bdpq

- gentler f

- new shape for r in keeping with pen angle, not sure I like it

- more narrow a

- two-story g

Next Day:

- spacing is wack

- fix r

- e, c feel clumsy

- another stab at a

- widening upper case

- more upper case glyphs

[Catherine Leigh Schmidt]

Focused on lower case today, plus a few details in the deva half-forms

Today:

- re-tuned a, e, c

- adjusting r, s

- kvwxyz

- stab at 123

- easy half forms of न, ग, ज, व etc and a few tricky ones

- a little bit of punctuation 

- spacing

Tomorrow:

- cleaning up lower case

- spacing lower case

- more latin numbers

- refining generated conjuncts

- upper case

There is a proof of the lower case in the repo! https://github.com/cathschmidt/yatra-one/blob/master/proofs/20151112yatra-lowercase.pdf

Two questions:

1. How does versioning work for fonts? How do you decide how to what counts as a new version?

2. As you can see from the proofs, I'm having some major issues with where the InDesign word-ready composer is sticking the matras. I'm not sure how this works and am looking into it. If anyone has any suggestions please let me know! Seems like the composer is putting matras everywhere without rhyme or reason — see the difference between मे & रे on page 4? What's up with that.

–c

[Dave Crossland]

On 13 November 2015 at 07:36, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

1. How does versioning work for fonts? How do you decide how to what counts as a new version?

Abstractly, we can distinguish between work that is public and publicised. Proprietary software/fonts are developed privately, and then tightly couple their releases' publication and publicity, marking clear versions in each release. Publicly developed libre software/fonts necessarily make more of a difference between the two concepts, since publication is constant. 

Practically speaking, the availability of fonts in the Google Fonts API is the primary point of publicity. A secondary point is the Github releases system -https://help.github.com/articles/creating-releases/ - which is a good way of marking new versions when you wish to. Its important that the version fields inside the source and binary font files in a release (eg in the NAME table, or Font Info inputs) match the version labelled on the release. 

That github help article links to http://semver.org and while this is a good version numbering scheme for software, since the binary font metadata field can only have one period separator, for fonts a MAJOR.MINOR-or-PATCH scheme is better, starting with 1.000 and incrementing from there (1.001, 1.002...) 

It would be good to have some note in the version string where possible like 'development version' that is removed when making a release build. 

 

2. As you can see from the proofs, I'm having some major issues with where the InDesign word-ready composer is sticking the matras. I'm not sure how this works and am looking into it. If anyone has any suggestions please let me know! Seems like the composer is putting matras everywhere without rhyme or reason — see the difference between मे & रे on page 4? What's up with that.

I suggest the 'y' might be better with a straight tail, also the j; the r arm is a little too wide. The spacing needs further work, the left side of t can be tighter, and pq seem very far apart. Might be worth chatting with Andrés Toresi (CC'd) about if he can apply his autospacer - https://www.youtube.com/watch?v=FrFGD3tzqig :)

[Khaled Hosny]

On Thu, Nov 12, 2015 at 04:36:04PM -0800, Catherine Leigh Schmidt wrote:
> 2. As you can see from the proofs, I'm having some major issues with where
> the InDesign word-ready composer is sticking the matras.

AFAIK, InDesign’s world-ready composer, uses a version (fork?) of ICU
LayoutEngine; and long outdated and currently deprecated OpenType
implementation that is known to have serious bugs. My suggestion is to
just ignore it and hope that Adobe will one day switch to a less buggy
implementation. Since Google Fonts are primarily for use on the web, how
web browsers render the font should be all that matter (fortunately
browsers tend to have much better and up to date OpenType support in
general).

Regards,
Khaled

[Dave Crossland]

I agree, I recommend testing OpenType features in browsers - primarily Android Chrome, iOS Safari (which may also not work well, depending on the script) and major desktop browsers :) 

[Catherine Leigh Schmidt]

Dave & Khaled, thanks for your feedback!

Dave — Thanks for all the info on versioning. Very helpful. I agree with j & y, and yes the spacing needs a lot of love.

Khaled — good to know about the word-ready composer. Will test in browser from now on.

I saw on another thread that someone had developed a nice browser proof. Anybody have a link to that? Will see if I can find it again...

[Dave Crossland]

On 14 November 2015 at 00:27, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

I saw on another thread that someone had developed a nice browser proof. Anybody have a link to that? Will see if I can find it again...

www.impallari.com/testing ?

[Catherine Leigh Schmidt]

Yep! Will also make another one to live in the repo.

[Erin McLaughlin]

Just looked at your file - the abvm.e mark on your Ra isn't in the right spot - that's why the mark was floating. You just need to move it down! :)

[Catherine Leigh Schmidt]

n00b move ;) thanks for spotting it!

[Catherine Leigh Schmidt]

Thanks Dave, Khaled, and Erin for the feedback! 

Today:

- added more angularity in the lower case (o, c, d, p, q, b)

- some spacing in the lower case

- redid y

- fixing e

- fixed plugged up knots in म भ

- more angular punctuation and anusvara

- upper case ABCHIPRQOSFE

- moved abvm.e & other anchors that were in wacky places

Tomorrow

- spacing

- remaining deva base chars

- upper case

- diacritical marks

- numbers

- smaller counters in lower case

- testing matra placement

proof is here: https://github.com/cathschmidt/yatra-one/blob/master/proofs/20151116yatra.pdf

Looking at this I think all of the spacing on the lower case is too loose. Anybody agree with me?

[Catherine Leigh Schmidt]

Today:

- more tuning in the lower case a e c g y

- some upper case CDGLM

- some sorts & punctuation ' " $^_-+=/\[]:;.,

- fi & fl ligatures...which seem a little superfluous but the glyphs are there so

- switched from round to diamond tittle, anusvara, period, etc...

- slightly tighter spacing on deva lower case

- more deva numbers

- bit the bullet for क्ष, other conjuncts, रु, रू

- some deva spacing but needs a lot of work

Tomorrow:

- fix plugged म, भ, ७ — contours in wrong direction I guess

- latin numbers

- evening weights in upper case

- spacing in upper case

- even tighter spacing in lower case

- investigating positions of marks....

Proof available here: https://github.com/cathschmidt/yatra-one/blob/master/proofs/20151117yatra.pdf

[Catherine Leigh Schmidt]

Sorry for the gap between updates — this is a big one

"Today":

- evening weights & shapes in upper case

- latin numbers

- math symbols

- diacritical marks, eth, thorn, ß, Æ etc 

- currency symbols, ®§¶™ etc etc 

- fractions

- spacing spacing spacing

Tomorrow:

- deva half-forms

- special conjuncts

- matra positioning

- ी's of varying lengths

[Catherine Leigh Schmidt]

Hi everybody! Big day today. Apologies for sporadic updates. Coming to the finish line here. The font is, by my estimation, mostly complete. All the glyphs have been drawn!! Would very much appreciate feedback.

Today:

- all glyphs drawn

- adjusting anchors to ensure proper matra placement

Tomorrow:

- spacing adjustments for deva

- hinting with ttfautohint

- cross-browser testing 

- proofs

- more anchor adjustment across all glyphs

I've got some specific questions for feedback:

1.

I'm curious about these conjuncts (द्म क्त्र न्त्र स्त्र), especially the "eyelash" ones, which I've never encountered. Should the न् + त्र and स् + त्र connect like क् + त्र? Probably all of the conjuncts should be looked at by a native speaker to ensure that they're legible.

2. How is the placement of the diacritical marks?

3. How should the deva numbers be aligned? To the top of the shiro?

I'll be putting up proofs shortly for evaluation.

Thanks,

Catherine

[Catherine Leigh Schmidt]

Hi everybody, long time no see.

Big thanks to Erin for giving extremely detailed feedback on my attempts at font-making. Have been working through her suggestions these past two weeks. Namely, the differences in weight and spacing between the Latin and the Deva. I noticed this earlier in the design process but was unsure how to go about it. At Erin's suggestion have been trying to thicken up the Latin in the least destructive way possible.

Today:

- Heavier weight in Latin caps and lower case

- fixed spacing of आ

- revisited घ, ग, ऋ, ज, च, ध, ल, ह

- changed loops of म, भ

- thickened and adjusted Deva numerals

Tomorrow:

- proofs

- heavier weight on Latin figures

- redoing several Deva characters

- checking unintentionally closed loops

- fixing rakar

- some missing glyphs?

- fix spacing of below-base characters

- move halant

- revisit some conjuncts

- kerning pairs

And ultimately:

- (auto) hinting

- figure out what's wrong with naming (the font is called "SomeDevanagari", conveniently).

Thanks everybody! Apologies about not including more proofs, I'll be cooking those up shortly.

[Catherine Leigh Schmidt]

Hey y'all, just working through the details here.

Today:

- heavier weight on Latin figures

- redoing several Deva characters

- fixing rakar

- fix spacing of below-base characters

- moved halant

- revisit some conjuncts

- kerning pairs

- revising some glyphs & half-forms

Tomorrow:

- check plugged loops (na conjuncts, १ as seen on proof...!)

- adjusting spacing on half-forms

- adjusting some conjuncts

- kerning pairs

- revisiting na conjuncts

- devanagari spacing, in general

Hoping that the weights of the latin figures, deva figures, upper and lower case, and devanagari feel more or less the same weight now. Will be checking the spacing of the devanagari with this http://motaitalic.com/devanagaritester/, which has been an absolutely amazing resource. 

Am realizing that I'm missing a few devanagari characters and I'm at a loss as how to add them without breaking anything.

A proof is attached here, and alternatively is available in the repo here http://github.com/cathschmidt/yatra-one/blob/master/proofs/20160121yatra.pdf

Many thanks,

Catherine

[Dave Crossland]

Thanks Catherine! I'll take a look and revert shortly

[Catherine Leigh Schmidt]

Hi everybody,

More detail work and quick fixes based on testing.

Today:

- fixing plugged loops & knots

- revised म भ knots (and all their conjuncts...)

- adjusting spacing on half-forms

- adjusting some conjuncts

- checking abvm.e and bindu mark placement...very wacky on a lot of conjuncts

- fixed rakar conjuncts

- fixed na conjuncts

- fixed rakar on द्र

- fixed placement of u matra on द

- revised shapes for ल, ज, च

Tomorrow:

- reviewing conjuncts with ल, ज, च, which all have revised shapes

- review spacing on half-forms (most should be tighter)

- adding bit of shiro to half-forms that require it, like थ, श, ध etc

- adding missing glyphs

- kerning pairs in latin caps

A few questions:

1. how do I add missing glyphs without breaking everything

2. I'm missing अॆ, which I probably need. I'm also missing ऩ and ऴ — should I add these? I'd also like to add the Marathi forms for श and ल since this project came out of Bombay signage, though maybe that can wait for another version.

3. Is there a half-form for ङ besides ङ्?

Thanks !

[Dave Crossland]

On 26 January 2016 at 14:59, Catherine Leigh Schmidt

<cathls...@gmail.com> wrote:
> 1. how do I add missing glyphs without breaking everything

Missing, against what? :)

Adding glyphs should not, by itself, effect any OpenType logic you have already.

> 2. I'm missing अॆ, which I probably need. I'm also missing ऩ and ऴ — should
> I add these? I'd also like to add the Marathi forms for श and ल since this
> project came out of Bombay signage, though maybe that can wait for another
> version.

I would like to see Marathi support, yes :) Ek Mukta may be a good
reference for features.fea code

> 3. Is there a half-form for ङ besides ङ्?

Erin? :)

[Catherine Leigh Schmidt]

Missing, against what? :)

 

Against... well, since this is directly on the ITF base glyphset, nothing, I guess. But it really does seem like a good idea to have अॆ, ऩ and ऴ no? Mostly gauging from the Impallari tester.

So I just added those glyphs plus Marathi ल श (I've had them stashed away), but I can't get the build script to work now that they're in. I'm hunting around various features.fea files from ITF and EkType and am having a hard time finding what I need. Plus I don't understand this build script well enough to make any changes without breaking anything.

I'm getting this error:

makeotf [Error] Failed to convert input font 'styles/Regular/font.ufo' to temp file 'styles/Regular/font.ufo.tmp' with tx.

tx converting to T1 font: <tx: --- styles/Regular/font.ufo

tx: (ufr) Warning: Encountered empty <string> for fontinfo key styleName. Skipping

tx: fatal error

Seems like doesn't have anything to do with new glyphs, except new glyphs are the only different thing about this. Also, I can't find styleName in anything :):):)

[not to mention that I can't get the name of the font correct, or get it to build it without interpolating...between two of the same file hahaha]

I assume the names of the glyphs need to be added in GlyphOrderAndAliasDB, so I'll try that tomorrow and see how it goes. And if anybody has any suggestions for code for Marathi substitutions to add to features.fea, that would be much appreciated :)

[Erin McLaughlin]

Hiya Catherine! Yep, अॆ is a part of the Devanagari unicode set, I believe. And you might as well put the other two in there, since they're easy!

Here's my quick answer to the ङ् question -- Other than the halant form, there may be vertically-stacked conjuncts that have ङ at the top of the stack, but unless you're constructing those via components, there's no other version of ङ् that you need as a stand-alone (that I know of, anyway!)

I will look at your progress tonight and write more comments here if I think anything needs attention!

I'll also try to see if I can figure out the issue you're having above.

Thanks,

E

[Erin McLaughlin]

(here are the ङ् combos in Ek Mukta, for reference)

[Erin McLaughlin]

Just sent Catherine an email with some comments!

[Catherine Leigh Schmidt]

Hi all,

Erin, thanks again for your feedback. Have been incorporating your insightful suggestions. The good news is I'm hoping this is mostly finished. The bad news is I've broken ITF's build script again, so there are no otfs with this commit.

Today:

- adjusted size of space

- adjusted weight and width of a few characters उ द्ृ छ श्र

- adjusted space on quotes, created kerning pairs for quotes and punctuation

- "neatened" some half forms

- refined shapes of i matras (hilariously, my autocorrect is giving me "mattress" for "matras")

- adjusted position of anusvara

- kerning pairs in upper case Latin

Tomorrow:

- getting an a real otf >:( for some more testing

- (auto)hinting

- correct naming

- getting the Marathi and other missing glyphs in without breaking anything

- (web) specimen

The break has something to do with the ufos being given, because nothing about the script has changed. Frustratingly I reverted to an older version of the font without new glyphs: this also breaks. If not new glyphs...then what? 

Like others on this group, the shaping features don't seem to work correctly in my version of InDesign. While I desperately would like them too, I'm ok if everything works right in a browser.

A few weeks ago I tried ttfautohint (both gui and command line) and received an error that the file couldn't be read. I'll try again as soon as I can get an otf.

Many charming errors. Certainly this is a learning process. If anyone has suggestions for helpful documentation, I would really appreciate it.

Thanks,

C

[Dave Crossland]

Hi

On 28 January 2016 at 20:51, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

Erin, thanks again for your feedback. Have been incorporating your insightful suggestions. The good news is I'm hoping this is mostly finished. The bad news is I've broken ITF's build script again, so there are no otfs with this commit.

hehe okay :) Please update the https://github.com/cathschmidt/yatra-one with the steps you take to go from saving /masters/Yatra_{0,1}.ufo to OTFs, so I can try to reproduce the build.

 

A few weeks ago I tried ttfautohint (both gui and command line) and received an error that the file couldn't be read. I'll try again as soon as I can get an otf.

ttfautohint only works on ttfs, so you'd need to convert the otf to ttf.

 

Many charming errors. Certainly this is a learning process.

https://twitter.com/Ziweeee/status/690981513719070721

If anyone has suggestions for helpful documentation, I would really appreciate it.

What FDK documentation did you read so far? 

I see I never did get around to publishing my FDK workshop notes from last year... I think I got blocked on figuring out what to do with the examples and stuff. Well, here's the README, and I'll ask Frank G at Adobe what's up.

https://github.com/davelab6/afdko-workshop-coopertype-2015

--

Cheers
Dave

[Catherine Leigh Schmidt]

Hi all, just updated the repo with the latest ufos which are breaking the build script.

These are the steps I take to compile the font:

1. I'm working in Fontlab Studio 5, so I use RobofabUFO to convert the .vfb into a .ufo

2. I put that .ufo into the /masters directory, make a copy, and name them Yatra_{0,1}.ufo

[The script can be run with --nointerpolate, but I've never gotten it to work. So, two files it is.]

3. In the command line I run "python build.py -grimc" which to the best of my knowledge generates OpenType classes, resets styles/instances, generates instances, matches i matras to base glyphs, and compiles otfs.

4. I get this error:

makeotf [Error] Failed to convert input font 'styles/Regular/font.ufo' to temp file 'styles/Regular/font.ufo.tmp' with tx.

tx converting to T1 font: <tx: --- styles/Regular/font.ufo

tx: (ufr) Warning: Encountered empty <string> for fontinfo key styleName. Skipping

tx: fatal error

Don't know where styleName is or how to put a valid string in there. 

The .ufos currently in the repo don't include the missing and Marathi glyphs; I got em, but I want to make sure this works first.

Currently going through the afdko workshop notes! 

As for ttfautohint, this makes sense. Now, how to go from an otf to a ttf? I'll be doing that today.

The other issue is naming, and I'll be poking around with that today too.

Thanks all,

Catherine

[Catherine Leigh Schmidt]

What's up everybody! I'm seeing light at the end of the tunnel!

After reading the afdko workshop notes and the OpenType Cookbook I feel like I have a slightly better grasp of this whole thing. With some fiddling, I think I have the naming bug sorted and the new glyphs are in. There's a little feature for Marathi support as well (just Sha and La). Also I figured out how to run the script without interpolation, haha.

Today:

- ff, ffi, ffl ligatures

- added Marathi glyphs

- added "Short A" and "Short E matra"

- fixed name

Tomorrow:

- १ is still plugged!!!! I'm obviously drawing it incorrectly but I'm not sure exactly how. I also don't quite understand path directions, so I'm going to look into that.

- convert to ttf and autohint with ttfautohint

- build a web specimen! woo!

From what I can tell the AFDKO also has an autohinter — any reason to use ttfautohint instead of that?

I think after this I will be d o n e !

[Dave Crossland]

Hi

On 1 February 2016 at 09:53, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

1. I'm working in Fontlab Studio 5, so I use RobofabUFO to convert the .vfb into a .ufo

Maybe better to use http://blog.fontlab.com/font-utility/vfb2ufo/
 

Don't know where styleName is or how to put a valid string in there. 

Its an element in the fontinfo.plist file inside a UFO: 

http://unifiedfontobject.org/versions/ufo2/fontinfo.html

How to go from an otf to a ttf?

You can do that, but its not idealy, because the OTF (CFF) format doesn't allow for components/references, and so when you convert an OTF to a TTF, the filesize is bigger than it has to be. 

Its better to compile the TTF from your source files directly. 

Cheers
Dave

[Dave Crossland]

Hi

On 5 February 2016 at 18:46, Catherine Leigh Schmidt <cathls...@gmail.com> wrote:

What's up everybody! I'm seeing light at the end of the tunnel!

:D

I have a print of your last test doc pdf that I have marked up, and will type up the comments asap :)

 

After reading the afdko workshop notes and the OpenType Cookbook I feel like I have a slightly better grasp of this whole thing. With some fiddling, I think I have the naming bug sorted and the new glyphs are in. There's a little feature for Marathi support as well (just Sha and La). Also I figured out how to run the script without interpolation, haha.

Great!

 

Today:

- ff, ffi, ffl ligatures

- added Marathi glyphs

- added "Short A" and "Short E matra"

- fixed name

Cool!

 

Tomorrow:

- १ is still plugged!!!! I'm obviously drawing it incorrectly but I'm not sure exactly how. I also don't quite understand path directions, so I'm going to look into that.

Sure. 

 

- convert to ttf and autohint with ttfautohint

ttautohint is only for post-processing already made TTFs.

 

- build a web specimen! woo!

:D

 

From what I can tell the AFDKO also has an autohinter — any reason to use ttfautohint instead of that?

The fdk autohinter is for OTF/CFF fonts

--

Cheers
Dave