Akaya : Kannada + Telugu + Latin

[vaishnavi Murthy]

A big hello to everyone on this forum. 

Its really exciting to see so many nice new typefaces being built this year.

I'm currently working on a Kannada + Telugu + Latin calligraphic display typeface called Akaya.  

A being the first vowel, 

KA the first consonant & 

YA the first semi-vowel 

Akaya also means -- older sister -- and is relatively easy to remember.

I plan to log my work progress here : https://github.com/vaishnavimurthy/Akaya/blob/master/Test-Files/readME.md#log-sheet

Do take a look. Don't hold-back on any form feedback (as long as it doesn't involve physical violence) :)

-V

[Liang Hai]

Really an exciting project! And a very nice name.

Btw, uh, what do you mean by "cluster breaking behaviour" and "automatic cluster breaking for complex uncommon subscript forms" (as written in /Test-Files/readME.md)? These two lines interest me. :)

梁海 Liang Hai

[vaishnavi Murthy]

There is a difference between क्क्क्क & ಕ್ಕ್ಕ್ಕ. The b.b

forms in Kannada form a huge gap in the text that 

doesn’t suit display typefaces. I’m not sure if I’m 

using the right words, but what I mean is this:

 द्ध् + घ  =  द्ध्ध . A way to define such behaviour.

The shaping engine currently allows for infinite

below base strings : క్క్క్క . The logic is  currently that 

you fix most of the commonly used forms and the

rest don't render well and that’s ok because

it’s rare. Maybe there’s a way to fix this. And becuase

I haven’t looked into this much so far, I wanted

to find-out how these issues are handeled. I’d love 

to know what the various approaches to this are

for Kannada and Telugu.

What I’d like to happen in Akaya is to restrict the 

below base forms to a maximum of 2 and 

break the cluster with a virama. This would be true

for both Kannada and Telugu:

క్క్క్క =  క్క్క్‌‌క

ಕ್ಕ್ಕ್ಕ =  ಕ್ಕ್ಕ್‌ಕ

There would be exceptions to this like క్క్షై.  

This would render with a b.b cluster of 3 because its 

common, but  క్చ్హై =  క్చ్‌హై . This would break because

it is rare. Is there an exhaustive corpus of such common 

clusters that exists? If it does, then the rest can be

made to break with a maximum limit of 2. This would:
(a) eliminate rendering issues & (b) for a more even texture 

on the page whenever possible.

-V

[Sarang Kulkarni]

Great Vaishanvi. Looking forward to see the font soon.

Best wishes

Sarang Kulkarni 

[vaishnavi Murthy]

Thanks Sarang! Do share your expert views/comments on this when you find time.

Hellos to everyone at Ek type & good luck with your massive type family ;)

-V

[Erin McLaughlin]

OH MY GOSH, beautiful shapes!!! Sorry I am late to reply!

Telugu's subscript stacking is so intimidating to me! Eek! Which fonts that exist today do you think actually deal with that subscript situation well? I'm just curious :)

[vaishnavi Murthy]

Thanks Erin!

I agree, Telugu subscript situation is pretty Eekish. I don't think there are many fonts that deal with this situation well. What do you think? Each font has its own version of dealing with vattus and matras--usually arising from the bounding box constraints. Text and display have two very different behaviours as-well.

I'll share a few notes on this soon. 

-V

[Liang Hai]

@Vaishnavi:

> There is a difference between … … … on the page whenever possible.

Personally I always think such control on consecutive C2 marks is supposed to be authors/editors' duty (it's basically a spelling issue, although it usually doesn't alter the expressed phonetic sequence). Text engine developers are the ones responsible and they need to ensure ZWNJ and ZWJ are effective and are easy to input.

However, it's should still be possible to hack OTL logic to produce the result you want, though it'll be a lot, a lot hacking, because of the reordering work done by OTL engines.

[vaishnavi Murthy]

> I always think such control on consecutive C2 marks is supposed to be authors/editors' duty... - - - ... still be possible to hack OTL logic to produce the result you want...

Aha! that is exactly what I wanted to know :) I'm trying to find some logic to make this happen---mainly to avoid garbled clusters with overlaps. Do you know of any such experiments with either Kannada, Telugu or Malayalam?

Thanks Liang.

[Liang Hai]

Nope, I don't know any of such. I've quickly tried some basic ideas (decomposing C2 marks back to <virama, C> sequences; preventing a C2 mark's forming in <akhn>; …) but they didn't work well because of the reordering.

The reordering is still possible with some heavy hacking, for example:

<KA, mAA, KAc2, KAc2, KAc2> == decomposing ==> <KA, mAA, KAc2, KAc2, Virama, KA> == multiple steps of swapping adjacent glyphs ==> <KA, Virama, KAc2, KAc2, KA, mAA>

> mainly to avoid garbled clusters with overlaps

Ligating and spacing (instead of re-organize the syllable structure) might even be easier and more reliable (and, more of a font developer's duty) if you only want to avoid "garbled clusters with overlaps".

[vaishnavi Murthy]

I see how you're distinguishing between a character and its representation. I'm sorry if what I meant was not clear. 

I've put-together a document for this. 

Since Akaya is a display typeface, It's much easier and the documentation is not really as complicated as a text face.

Do take a look when you have time. I'd love to get some feedback on this as I'm sure there are several corrections to be made to it and maybe I'm wishing for too-much :)

Best,

Vaishnavi

On Monday, 24 August 2015 Liang Hai wrote under Hind thread: 

I'm confused by what you meant by "horizontally spacing below base forms (and the following length mark in case of Kannada)"…

Do you mean 1) <KA, KAc2, KAc2, KAc2>, or 2) <DA, DAc2, DAc2, DAc2>?

Personally I say "C2 mark" to describe a glyph's writing/orthographical status, while "pre/above/below/post-base" to describe a glyph's visual characteristics.

Therefore,usually Telugu KAc2 (the C2 mark of KA) is a spacing post-base C2 mark, while Telugu DAc2 is a non-spacing below-base C2 mark, while Kannada Lengthmark is a spacing post-base vowel mark.

Then, "horizontally spacing below base forms" means … — do you mean the alternative spacing glyph variants of non-spacing below-base C2 marks? Then Murty Telugu's extensive GPOS adjustment is a good reference to what's possible.

But I guess you prefer Telugu's traditional stacking behavior for below-base C2 marks?

Anyway, , it'll be easier to talk about when there's some specific case.

[Liang Hai]

Fake small-caps are neither legible or pretty or friendly to copy-n-paste…

## PBF behaviour

1) Treat క్ష as an atomic unit (a letter, instead of "a letter + a PBF") then the logic will be more clear.

2) Interesting. I thought it was like that only when it's in క్ష, turned out it's actually ecceptable everywhere?

3) … This is one of the reasons why I don't use "post-base" to describe a C2 mark. Your T.PBF and V.S.PBF RA could be better described as "a pre-base C2 mark of RA" vs "a below-base C2 mark of RA". I guess "when its the second PBC in a cluster" could be better put "when it's not the first PBC"?

a) Awesome! Finally I found the proof of AI Length Mark's special relationship with RAc2. Will fix this issue in Kohinoor Telugu. Again, if you treat క్ష as an atomic unit you don't need to declare క్ష్ర a special case.

b) Wouldn't it be easier if you treat క్క్ష as "KA + atomic KSSAc2" (instead of "KA + KAc2.alt + SSAc2") then move "atomic KSSAc2" into your conditions of 3.a? (Although it's against my understanding to see you still keep the V.S.PBF RAc2 when it's under a deep KSSAc2.)

## VM AI placement with PBF

4) I though you'd use the old PBF of MA here.

5) Interesting.

6) In this phonetic sequence I feel, ర్‌ఙ్గై would a better spelling? It's a bit awkward to break between ఙ and గ. (This is also one of the reason why I feel it's hard to break consecutive multiple C2 marks properly, since it'd be often bad to break "between a same-class nasal and stop" or "between a stop and a semi-vowel", and it's even dependent on the etymology of a word to choose a good breaking way.)

a) Again, try treat క్ష as an atomic unit. Also, classify TAc2 and the old MAc2 into some special "thin below-base C2" group.

b) RA's below-base C2 mark (or in your terminology, RA's V.S.PBC) should be classified into that "thin below-base C2" group too.

## Cluster formation

7) …

a) Some ambiguity: I guess you mean it should be placed below *both* the base and the first PBF while horizontally aligned to the base (as shown in your example syllable)? Ahh, I thought TAc2 should be placed simply under its preceding (tall) post-base C2 mark instead also "horizontally aligned to the base".

b) This rules produces conflicts with 3.b, 6.a.(eg), 6.b. (Reconsidering how to treat క్ష related stuff seems necessary.) Also, what do you mean by "by adding ZWJ after the halanth"? Beaking consecutive C2 marks uses only ZWNJ, isn't it?

## Special marks behaviour

8) In the "Note", your character sequence for Reph is wrong.

9) Do you mean, in న్ద <NA, Virama, DA>, to enable Nakara Pollu you expect <NA, ZWJ, Virama, ZWNJ, DA>? Have you considered simply using the Stylistic Sets <ssXX> feature or the Stylistic Alternates <salt> feature?

## RESOLVE

10) Looks like KAc2's alternative form? ష్క్రాం sounds correct and it's mentioned in Edward C. Hill's "A Primer in Telugu Characters" too. (Well it's basically an alternative form to that mark in your section 11 issue.)

11) I think it's usually considered simply an alternative form of KAc2. (It's kind of a parallel behavior to MAc2's two forms.) Well, early typesettings could be very random…

12) Considering Telugu and Kannada's common origin, it's really hard to tell what a "borrow" is if it happened before these two scripts' modernization. And this piece of print, itself appears to be a very Kannada-styled Telugu, considering that లీ/ಲೀ on the beginning of the first line and that శొ/ಶೊ in the third line as well as those curly E vowel marks…

13) Mmm, since it's a display face, it's really not that important to limit vertical metrics…

14) Ah… So it's not a myth for me to think the blue one is sort of more standard? Or by "widely accepted" you only mean "common in fonts" instead of "common in people's daily writing and lettering"?

15) I though you don't plan to use such side-by-side fallback. If you do, then Murty Telugu's efforts on this is the best you can find.

16) What do you mean by "when you have space for 3 v.s.pbf anayway"?

[Vaishnavi Murthy]

Wow! That's some really meticulous feedback! 

 

## PBF behaviour

1) Treat క్ష as an atomic unit (a letter, instead of "a letter + a PBF") then the logic will be more clear.

Well, the logic here will apply to other strings as-well where : a letter + SSA PBF is present and not just with KA. For example SSA+Halanth+SSA+Halanth+MA will have the vertically stacking MA post-base form.

 

2) Interesting. I thought it was like that only when it's in క్ష, turned out it's actually ecceptable everywhere?

It has been that way. But I'm choosing to keep it the same because I'm defining the behaviour specifically for the typeface I'm working on and what I'm working on is very calligraphic. 

 

3) … This is one of the reasons why I don't use "post-base" to describe a C2 mark. Your T.PBF and V.S.PBF RA could be better described as "a pre-base C2 mark of RA" vs "a below-base C2 mark of RA". I guess "when its the second PBC in a cluster" could be better put "when it's not the first PBC"?

 

I'm sorry Liang, I've lost you here :) 

 

a) Awesome! Finally I found the proof of AI Length Mark's special relationship with RAc2. Will fix this issue in Kohinoor Telugu. Again, if you treat క్ష as an atomic unit you don't need to declare క్ష్ర a special case.

Great! Same thing here, it'll be restricting SSA to KA combination.

 

b) Wouldn't it be easier if you treat క్క్ష as "KA + atomic KSSAc2" (instead of "KA + KAc2.alt + SSAc2") then move "atomic KSSAc2" into your conditions of 3.a? (Although it's against my understanding to see you still keep the V.S.PBF RAc2 when it's under a deep KSSAc2.)

I agree with you here. KSSAC2 will make sense as an atomic unit here. This tall KSSAc2 combination is specifically used that way.

 

## VM AI placement with PBF

4) I though you'd use the old PBF of MA here.

Me too haha! I was very caught-up looking at the AI here maybe. Will fix it. Thanks.

 

5) Interesting.

6) In this phonetic sequence I feel, ర్‌ఙ్గై would a better spelling? It's a bit awkward to break between ఙ and గ. (This is also one of the reason why I feel it's hard to break consecutive multiple C2 marks properly, since it'd be often bad to break "between a same-class nasal and stop" or "between a stop and a semi-vowel", and it's even dependent on the etymology of a word to choose a good breaking way.)

I chose the example particular because its a cluster with a bit of an challenge when it comes to breaking : a semi-vowel and nasal stop combination. I still need time to resolve this issue.

 

a) Again, try treat క్ష as an atomic unit. Also, classify TAc2 and the old MAc2 into some special "thin below-base C2" group.

b) RA's below-base C2 mark (or in your terminology, RA's V.S.PBC) should be classified into that "thin below-base C2" group too.

 

Making a "thin below-base C2" group sounds like a good idea. 

 

## Cluster formation

7) …

a) Some ambiguity: I guess you mean it should be placed below *both* the base and the first PBF while horizontally aligned to the base (as shown in your example syllable)? Ahh, I thought TAc2 should be placed simply under its preceding (tall) post-base C2 mark instead also "horizontally aligned to the base".

Yes, it should be placed under the tall consonant and also align horizontally with the base character.

 

b) This rules produces conflicts with 3.b, 6.a.(eg), 6.b. (Reconsidering how to treat క్ష related stuff seems necessary.) Also, what do you mean by "by adding ZWJ after the halanth"? Beaking consecutive C2 marks uses only ZWNJ, isn't it?

This is a general rule that needs to be followed with expections of the specific rules mentioned above.

Please ignore the ZWJ remark. Its my personal notes I forgot to delete. I was seeing how to allow unusual combinations one occasionally finds in modernist poetry. Am still thinking about it.

 

## Special marks behaviour

8) In the "Note", your character sequence for Reph is wrong.

Fixed it. Thanks!

 

9) Do you mean, in న్ద <NA, Virama, DA>, to enable Nakara Pollu you expect <NA, ZWJ, Virama, ZWNJ, DA>? Have you considered simply using the Stylistic Sets <ssXX> feature or the Stylistic Alternates <salt> feature?

It is pretty crazy isn't it? haha! I want to use it as a stylistic variant. Thanks for pointing it out.

 

## RESOLVE

10) Looks like KAc2's alternative form? ష్క్రాం sounds correct and it's mentioned in Edward C. Hill's "A Primer in Telugu Characters" too. (Well it's basically an alternative form to that mark in your section 11 issue.)

Well, I think its a badly printed KA PDC here. I agree that Edward C Hill's primer is a great reference but I know this mark is also sometimes used as a length mark. I'm not including it for this typeface but it was more a note for me. Can ignore it for now maybe.

 

11) I think it's usually considered simply an alternative form of KAc2. (It's kind of a parallel behavior to MAc2's two forms.) Well, early typesettings could be very random…

Haha, early typesetting is fun! Take a look at the MAc2 form here

 

12) Considering Telugu and Kannada's common origin, it's really hard to tell what a "borrow" is if it happened before these two scripts' modernization. And this piece of print, itself appears to be a very Kannada-styled Telugu, considering that లీ/ಲೀ on the beginning of the first line and that శొ/ಶೊ in the third line as well as those curly E vowel marks…

Kannada and Telugu are very similar I agree. This alternate form is mentioned in a couple of places. One of them is also in Hill's primer. Once again I just want to be sure that its correctly documented.

 

13) Mmm, since it's a display face, it's really not that important to limit vertical metrics… 

14) Ah… So it's not a myth for me to think the blue one is sort of more standard? Or by "widely accepted" you only mean "common in fonts" instead of "common in people's daily writing and lettering"?

It certainly is common in fonts. Writing styles vary a lot as you know and therefore can't really comment on it.

 

15) I though you don't plan to use such side-by-side fallback. If you do, then Murty Telugu's efforts on this is the best you can find.

No I don't plan to use that. But I'm looking at stylistic sets maybe. But it sounds very complicated for the time being. So, it'll either have to be the vertical placement or horizontal depending on what can be achieved with least amount of errors.

 

16) What do you mean by "when you have space for 3 v.s.pbf anayway"?

This is addressing what I'd defined under (7.).

-V

 

[vaishnavi Murthy]

Work update :

https://github.com/vaishnavimurthy/Akaya/blob/master/Test-Files/readME.md#30-aug-2014

-V

[vaishnavi Murthy]

Update : https://github.com/vaishnavimurthy/Akaya/blob/master/Test-Files/readME.md#7-sep-2014

-V

[Liang Hai]

Wow, wow it's awesome.

[vaishnavi Murthy]

Thanks Liang!

Here's the latest update :

https://github.com/vaishnavimurthy/Akaya/blob/master/Test-Files/readME.md#29-sep-2015

Will be working on the rest of the composite glyphs for both Kannada and Telugu next.

-V

[vaishnavi Murthy]

Update for the last 2 weeks.

- Tel. composite glyphs for combining vowel marks - all

- Tel. pre-base and below base marks. 

- Spacing Basic Kan. and Tel. to test GPOS

- 2 experiments on matching Latin 'adhesion' text.

- A basic working GPOS and GSUB feature file with : GPOS for Telugu base glyphs and GSUB for Tel. halanth forms and composite glyphs.

I'm working on Telugu first as its a more complex script than Kannada and will help with setting-up guidelines that'll roughly work for Kannada as-well.

I'll be uploading the files and test documents on Github as soon as I find better connectivity that'll allow for this.

Next :

TEL. post-base characters. 

Kannada composites.

[vaishnavi Murthy]

Update :

Working on Kannada composites for the next 3 days.

[Liang Hai]

> Working on Kannada composites for the next 3 days.

Great! This (how these marginal consonant-vowel ligatures are handled by you) is the part I've been looking forwards to the most!

[vaishnavi Murthy]

Yes, NYA and NGA v.ligatures are a bit tricky since they are a recent addition to the script to support Sanskrit. 

Kittel's grammar is taken as a standard for this. Not that its well resolved there.

[Juan Luis Blanco]

Hi everyone!

I am really pleased to have the opportunity to contribute to this project. Thanks Vaishnavi :-)

My task, challenging indeed, will be to design a Latin companion for the amazing Kannada and Telugu versions of Akaya. The first attempts show how challenging it is going to be. As a first presentation I would like to share with you a bunch of characters we are working on to define the overall feel of the Latin. In the images below you will see along with some previous drafts, two different proposal I named v00 and v01. They seek for the same flow, speed and particular construction of its Kannada and Telugu counterparts. Any feedback will be very welcome!

Juan

[Dave Crossland]

The letters' left tilt seems strange to me; wouldn't a right tilt make more sense for the casual, quite fast feeling?

--
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/8590ffc0-55bb-4973-9fa4-d93860d9458c%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Juan Luis Blanco]

Thanks for your replay Dave.

Do you mean the left tilt of the ascenders? The reason for that is that I was experimenting with an inverted ductus, trying to imagine what the letters would look like if they were written the other way round (this has proved much more difficult in some letters than in others). This reversed ductus is actually what produced such dynamic and fluid strokes in Kannada and Telugu, as Vaishnavi told me. The result are pretty unconventional shapes. I was looking for the same effect in the Latin script. However the feel of speed is probably going to be faster tilting the letters to the right although the result may end up being more conventional. I need to weigh that up.

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/8590ffc0-55bb-4973-9fa4-d93860d9458c%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Khaled Hosny]

On Sun, Oct 25, 2015 at 01:37:29AM -0700, Juan Luis Blanco wrote:
>
> Hi everyone!
>
> I am really pleased to have the opportunity to contribute to this project.
> Thanks Vaishnavi :-)
>
> My task, challenging indeed, will be to design a Latin companion for the
> amazing Kannada and Telugu versions of Akaya. The first attempts show how
> challenging it is going to be. As a first presentation I would like to
> share with you a bunch of characters we are working on to define the
> overall feel of the Latin. In the images below you will see along with some
> previous drafts, two different proposal I named v00 and v01. They seek for
> the same flow, speed and particular construction of its Kannada and Telugu
> counterparts. Any feedback will be very welcome!

I like the v00 one most, except for the d as it looks like the partial
differential symbol.

Regards,
Khaled

[Juan Luis Blanco]

Thank you for your feedback. I am aware of the similarity between d and partial diferential. I'll bear that in mind as the drafts evolve and develop.

[Juan Luis Blanco]

After drafting some more characters in both versions (v00 and v01) made a decision for v01. Some more lowercase letters have been added and everything has been refined. Github repository: Akaya_L_v011.glyphs

[vaishnavi Murthy]

Hello,

Attached are 2 separate pdf files with Kannada and Telugu composites. I'd love to hear some feedback for this.

Glyphs and TTF files along with basic feature files are uploaded here :  https://github.com/vaishnavimurthy/Akaya

Once again, any feedback regarding this will be welcome.

Update (Kan/Tel) : 

- All composite glyphs for Kan/Tel

- Basic spacing for Kan/Tel characters

- Set kerning groups for Kan/Tel

Next:

- Kerning Kan/Tel

- Kannada Below base marks and fix Tel Below base marks

- Build feature files

Thank you.

Best,

Vaishnavi

[vaishnavi Murthy]

UPDATE - K / T 

 

- - - - - - - - - - - - - - - - - - - - 

 

⤫ Kerning Telugu ⤑ Small fixes to base characters ⤑ Re-defining kerning groups

⤬ Cleaning Telugu Vattu 

⤫ Re-looking at Vattu combinations & their behaviour

[Juan Luis Blanco]

Latin lowercase completed. Changed d and f to match the rest of the letters. Refined most to them. Basic punctuation and first uppercase letters drawn. 

File: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v014.glyphs

[vaishnavi Murthy]

Update :

Akaya Telugu Kerning Overflow issue : https://forum.glyphsapp.com/t/kern-overflow-issue/3325?u=vaishnavi

Next :

Akaya Kannada Kern

[Juan Luis Blanco]

V0.015 includes first version of upper and lower case sets. I still have doubts with some characters (A, Y, N...). The uppercase characters may be a bit overweighted. File here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v015.glyphs

Sample:

[vaishnavi Murthy]

I'm facing more trouble with kerning. When I save a copy of the glyphs file. I see the kerning values are changing and for some reason they are listed under the wrong groups. Ex., image below. Also, there are some kerning values that are changing. Has anyone faced this issues before? Do let me know if you have any solutions to this. I'm currently using GlyphsApp 2.0

Thanks!

[Erin McLaughlin]

Hi Vaishnavi! I haven't used Glyphs yet, so I am unfamiliar with their kerning tools and wasn't able to figure out how to look at your strategy. I might try later tonight if I have time. The Adobe Kannada project also had about 2000 kerning pairs (due to the insane amount of subscript kerning we employed), but we didn't have an overflow issue. I've never had one and don't know how to deal with it, so I'm sorry that I am not sure how to help! 

But if you're still having issues, I wonder if what was said on the Glyphs message board is relevant - you might have to make more generalized kerning with fewer exceptions. If you require kerns for more than 1/2 of your glyph pairs (not counting subscript kerns), that's a sign that the spacing should be re-done. Sorry that I can't be of more help at this time! I'll try to keep looking at it this week!

E

[Juan Luis Blanco]

Work update: 

- weight of upper case letters adjusted. 

- new design for A.

- J, V, W, v, w and y modified. 

- First sketches of numerals.

File: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v016.glyphs

[Juan Luis Blanco]

Work update:

two sets of figures –ols style and lining–, superiors and fractions added today

File: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v017.glyphs

[Dave Crossland]

On 10 November 2015 at 06:42, Juan Luis Blanco <txo...@gmail.com> wrote:

- new design for A.

I like it but I wonder if a design that leans to the right (as VW do) may work better :) 

[Dave Crossland]

On 9 November 2015 at 17:24, vaishnavi Murthy <vaishna...@gmail.com> wrote:

I'm currently using GlyphsApp 2.0

2.0 is now pretty old, I recommend using the very latest version available. 

[Vaishnavi Murthy]

Sorry I meant Glyphs V. 2.2.1(819), referring to the V.2 format in general. I'm redoing the kerning from scratch currently in a more stable env., which is V. 1.4.5 (614). There were way too many issues with the newer one.

--
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, 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/CAEozd0wHPmm14NhLzsnJczmTVE5qQwh-ek7X8HQQosDu_rND1Q%40mail.gmail.com.

[vaishnavi Murthy]

Hi Dave,

It would be interesting to take a closer look at the regional interpretations the informal styles while matching scripts. 

The way I’m looking at it is this : Akaya Kannada/Telugu was directly inspired by existing hand-drawn display styles. As you know, these scripts too lean forward in several hand-written styles, very similar to Latins. Kannada and Telugu also have connected script styles which are rarely used these days. Its called Modi (not to be confused with the Gujarati based Modi… or our Prime Mister Modi who is also Gujarati). 

These right leaning hand written styles might be a direct influence of the huge number of cursive writing books we had to fill as kids in schools. I see this influence often in South Indian manuscripts post 1800s.

Akaya is mostly upright with a tendency to lean right but does the opposite at times (knTA). The style is informal and script-like with a rhythmic calligraphic undertone. Each letter behaves as a unit — the speed within a letter varies unlike a conventional script typeface where there is a definite rhythm. This varying speed determines the incline of the letter and is therefore slightly uneven. Latin too follows this at times. You see the letters are leaning to the right slightly more in Latin already becuause of the nature of this script. 

In Kan/Tel there are narrow quicker strokes followed by slower rounded, drawn-out ones—within a letter—balancing each letter as a unit. A similar interpretation applied across all letters make it work. The proportions are relaxed with comfortable counters.

I feel the Juan’s Lain speaks the same language as Kannada & Telugu within the Latin framework. 

Do let us know if you’re still not convinced :)

Best,

Vaishnavi

[Dave Crossland]

On 11 November 2015 at 17:49, vaishnavi Murthy <vaishna...@gmail.com> wrote:

Do let us know if you’re still not convinced :)

I'm convinced! :D

[vaishnavi Murthy]

Reposting a thread discussing a few issues I was facing while making Akaya :

I'm following the Hindkit naming convention as you'd suggested. Hindkit didn't make much sense to me when I began. Now I understand to some extent what it is and its useful to have the gsub tables all listed but I had to tweak it to make some of it work like under what feature the glyph substitutions are listed etc., and the glyphs that are included in Akaya. Pria was very patient in explaining how the gsub gpos etc work and she's been great in answer the most ridiculous of questions :) I still feel I have a lot more to learn, however, I get the underlying logic. 

Since each typeface has its own unique behaviour, I feel most designers I've spoken to prefer to build things from scratch and not use pre-built stuff just to keep track of what's happening at every stage of building the font. Sometimes going over the pre-built stuff takes lot longer as you know. 

Thinking out loud, maybe, if building a font can be broken into specific steps with things to update and check on completion of each stage might allow for a definite route to take for people like me. For example :

-- Identifying the gaps that exist in styles for various scripts and reference visual materials that can be pooled together.

-- A simplistic over-view of how type-design and font engineering works together.

-- A repository of useful python scripts categorised under various broad topics where they can be used.

-- A user-friendly OTF features manual which also incorporates peculiarities of each script and the language systems involved.

___________

-- Defining character-sets for styles. Ex : standardised/simplified, extended with vedic support, extended with archaic etc.,. Defining the behaviour of combining forms for each of these (this involves a lot of work).

-- A file containing all the required glyphs with the right naming conventions and guidelines for setting-up the metrics.

-- Having glyph orders that let you shuffle around the characters for testing.

-- Test documents (exhaustive) for spacing, kerning & testing on completion of each stage (based on the character groups and complexity of behaviour of each script.) 

-- Feature files thats more like a lego block that have tables that can be copy pasted as required. I see how Hindkit address this to a large extent.

I spent a lot of time trying to get an overview of all the stages involved as I wasn't very sure about how to approach this. If I didn't have a person like Pria who I could contact directly, it would have made this process a lot harder.

I'm sure you're looking into all this, but I thought I'd share a few things that I thought of while working on this.

Best,

Vaishnavi

On 11 Nov 2015 5:10 pm, "Dave Crossland" <da...@lab6.com> wrote:

On 11 November 2015 at 18:33, Vaishnavi Murthy <vaishna...@gmail.com> wrote:

4. Inexperience and lack of general clarity as to how one goes about making a functional typeface :) I'm surprised by how much more there was to learn...

Was hindkit or any other project helpful for this?

---------- Forwarded message ----------
From: Vaishnavi Murthy <vaishna...@gmail.com>
Date: 11 November 2015 at 17:03
Subject: Re: My visit to Bangalore
To: Dave Crossland <da...@lab6.com>

Causes for delay:

1. Finding a good method to simplify elaborate shapes. Trying to fit these shapes within the nearest group bracket.
2. Kerning and spacing base glyphs and composites that have a lot of overhangs. I'm still working on this.
3. Trying to understand how these fonts function technically
4. Inexperience and lack of general clarity as to how one goes about making a functional typeface :) I'm surprised by how much more there was to learn...

Overall its been really awesome to be able to understand all this. Hopefully it'll translate well by the end.

Thanks again!

Best,
Vaishnavi

On 11 Nov 2015 4:49 pm, "Dave Crossland" <da...@lab6.com> wrote:

Hi

What do you think the causes of the delay are? :) 

 

[Ambarisha Darbha]

Hi Vaishnavi,

I'm sorry that I'm sending few suggestions on Akaya Telugu lately. These are very delayed comments / suggestions. I apologize for that. You can go through and feel free to get back to me on these.

Regards,

Appaki Ambarisha D

[vaishnavi Murthy]

Dear Ambarisha,

Thank you for your reply. Its always good to get feedback from someone who is actively using the script :)

I'll take your advice and make the top bit of OO combining marks curve more to the right. Everyone who has seen this has given me that feedback.

I see how the అ, ఆ, ఓ & ళ might make you uneasy but as you know this is a stylized script style. I've been discussing this with a few sign board/election poster painters in Hyderabad (the kinds who use this style often and they feel this is ok. Maybe I'll show it to a few more people before taking a decision on this. If you have more feedback on these shapes, it'll be great!

Also, in my previous PDF there are a lot of archaic ligature forms (because I felt they were closer to the old hand written forms and existed as alternates). They won't be displayed in the default setting. I've attached a PDF here to avoid this confusion and clarify this further. The previous PDF was for looking at the stroke behavior and I'm sorry about this confusion. The PDF has a few errors in mark placement. I'm still working on this.

The ఔ top line you've consistently marked as having to be extended to the left. I was told, in hand-writing styles, the behavior as seen in Akaya is more common. Once again, it'll be good to know more about what you think here.

I do have a doubt on the corrections you've marked for ఝ. I don't see this much elsewhere :

Thanks again!

Best,
Vaishnavi

[Appaji Ambarisha Darbha]

Dear Vaishnavi,

I agree with most of your comments. అ, ఆ, ఓ if you keep them as it is no problem. But my suggestion will make them look better. Most of these changes might not be completely necessary, but just something to think about. But I feel ళ looks better if you make that change. I’ve not seen that extended anywhere. Archaic ligature forms are there in usage by some people. Even ఝ formats you mentioned are also. But in archaic forms the second for OO should not have that deergha on the end. My model is currently using by max people for the last two decades. I can send samples from current DTP (ASCII) packages which are popular here.

With Best Regards,

Appaji Ambarisha Darbha

mobile: 9177994407

--
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, 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/15346175-d1cd-40f1-92ed-e3042fcfdf1a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

<Akaya_Telugu_20151107.pdf>

[Vaishnavi Murthy]

Aha! Thanks for this. It'll be great if you could take a look at the PDF attached in my previous mail and let me know if you find any archaic forms. Thanks again.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/C4840DE3-5197-4A12-83CA-C5277515C409%40gmail.com.

[Vaishnavi Murthy]

Hi Erin,

I seem to have missed your reply. I'm re-looking at it now. Thank you !

Best,

Vaishnavi

--

You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, 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/ab2fcef7-0ff5-43d1-9a0d-38cd769f553c%40googlegroups.com.

[Juan Luis Blanco]

Punctuation added to the Latin version.

See PDF for a sample of punctuation inside text.

File: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v018.glyphs

[Juan Luis Blanco]

Hi everyone,

Akaya latin has currency and mathematic symbols now. You can have a look in the PDF attached or check the file at: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v019.glyphs

Best,

Juan

[Juan Luis Blanco]

Hello everyone,

I added diacritics and all the accented characters present in Adobe Latin 3 set to the Latin version of Akaya.

Here is the file: https://github.com/Txonliz/Akaya-Latin-font/blob/45cc14a90f0fdf44f69447db256884b6091c3b5c/Akaya_L_v020.glyphs

Attached you'll find a sample in PDF format.

Best,

Juan

[vaishnavi Murthy]

Hello,

Akaya Kannada update: A Test file is attached here with a few samples of Juan's beautiful Latin along with Kannada. 

Source files are uploaded here: https://github.com/vaishnavimurthy/Akaya

The rare combinations have not been paid much attention to avoid excessive kerning. 

A few combinations are breaking and this will be fixed soon. 

Any feedback/review at this stage will be very useful.

Thank you.

Best,

Vaishnavi 

[Dave Crossland]

This is great stuff :D 

[梁海]

Wow it's so cool!

Mmm seems there're some Telugu pieces mixed between Kannada ones… (SAME WIDTH SEQUENCES, VOWEL SIGNS…)

What is the widely accepted structure of JHO? Should it be like MO and YO (one less stroke compared to the plain consonant), or like other letters (without special behavior)?

InDesign's built-in Unicode database is too old, therefore it doesn't recognize the Kannada nukta. (Same for RRA and LLLA.) Don't worry. Just use Core Text or HarfBuzz etc to test nukta.

Latin looks very wide to my eyes, and they're larger than I'd expect.

Will it be nice to use the "ω" form of handwritten "w"? (Similar for "v".)

I find a more cursive style of Latin might go better with the Kannada/Telugu, however I bet you guys must've chosen this style for reasons.

Didn't get time to follow the whole thread. Sorry if I talked about something that is already discussed above… :)

--

You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.

Visit this group at https://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/8d77cefc-9d1c-44df-92fe-b2be2e3aa16a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[vaishnavi Murthy]

Thanks Dave.

Hi Liang,

You're right, the same width sequence is in Telugu. My mistake. Will delete that bit.

Thanks for the info about Nukta. Very useful to know.

The Latin style discussion happened earlier :) https://groups.google.com/d/msg/googlefonts-discuss/8GhUoeMZy-0/NcZWvdUgEgAJ

I'm quite happy with the matching, Kannada is wide too. Juan is working on the Latin. Its alway great to get feedback. 

JHO, MO, YO issue is discussed on Page 3 here : https://groups.google.com/d/msg/googlefonts-discuss/U1b0H81xt18/9piBEZ8GCwAJ

[vaishnavi Murthy]

Hello,

I have a technical question for the font engineering experts.

Akaya Telugu has a lot of overhangs and such and requires a lot of kerning. 

The base glyphs have be re-drawn to eliminate this as-much as possible and the basic spacing has been optimised. 

Despite this, there is a lot of kerning required for style and will not function ok without the kerning.

The trouble is, when I add a combining below base form, the kerning goes away. 

The attaching characters are already defined as marks and are non-spacing with zero width glyphs. 

Is there any way to retain the kerning of the sequence of base glyphs when this happens? Example below:

The glyphs and TTF file are attached here.

Thanks.

Best,

Vaishnavi

[梁海]

@Vaishnavi:

Your "tlTAc2"'s "GDEF.GlyphClassDef" value is still "2" in the TTF file. Once I changed it to "3", your "lookupflag IgnoreMarks;" in "dist" works as expected.

Right now even "గ్గ్" doesn't work as expected.

You might need to manually specify "GDEF.GlyphClassDef" values instead of letting makeotf generate it automatically. See https://github.com/itfoundry/hind-guntur/blob/master/features/tables.fea for renference.

--

You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/9bc3538a-5cab-44f5-acb2-129a7abddc41%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[Appaji Ambarisha Darbha]

Hi Vaishnavi,

To retain the the below base form’s position with Base Glyph you are using kerning. Right?

I don’t have any idea on GLYPHS software. But in FontForge you can use ANCHORs instead of this. Define Anchors in FontForge. Use them to position your "below base form” in tune with Base Glyph.

See the attached screenshots.

With Best Regards,

Appaji Ambarisha Darbha

mobile: 9177994407

-- 
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/9bc3538a-5cab-44f5-acb2-129a7abddc41%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

<Akaya_Telugu_2.52.glyphs><Akaya_Telugu_16-Regular.ttf>

[Vaishnavi Murthy]

Thanks Liang and Ambarisha for your replies. 

Its very useful to know this. Thank you thank you :) 

I was just wondering if there is a way to define this on Glyphs app without writing-out the extensive GlyphsList XML file.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/6AAFA3E7-386A-4C94-A6F9-F19613A56607%40gmail.com.

[梁海]

@Vaishnavi:

Not sure what you mean by "the extensive GlyphsList XML file", but, uh, no one's saying you need to write any XML file…

The problem you're facing is because of your "GDEF" table not being controlled properly, but you're not using Glyphs.app for OTL feature compiling anyway (at least I don't see any OTL code in your .glyphs file), why do you need to control "GDEF" in Glyphs.app?

I've seen your "GDEF" code in https://github.com/vaishnavimurthy/Akaya/blob/master/Features/Akaya_02.fea

It should work, as long as this piece of code is really incorporated into your font. (The current situation is clearly that your whole "GDEF" table is generated by MakeOTF itself.)

I'm not sure about your workflow though.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAO8-OAPKu1AU8b7dHWHYbU_mUWU8zS1bN2%2BHBa9MUErmhGpnbA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[vaishnavi Murthy]

Hi Liang,

I've just updated the files on Github. The one you were referring to was pretty unresolved and work in progress. Pria has been very patient and is helping me sort this out. The workflow I'm following right now is similar to Bandipur.

Yes, the GDEF is generated by default as of now. I'll update this as I attempt to polish this file further. Thank you once again.

Best,

Vaishnavi

[Juan Luis Blanco]

Hello everyone,

I completed all the symbols required for Adobe Latin 3 character set (see attached PDF).

Glyphs file here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v021.glyphs

Best,

Juan

[vaishnavi Murthy]

Hello,

I'm finding it hard to wrap my head around writing the GSUB table for Telugu.

From what I've tried so far, there's still are a lot of things that don't seem to be working ok. Please ref. the attached PDF.

All source files are attached here as-well.

Any advice to get this to work will be super useful :)

Thanks.

Best,

Vaishnavi

[Juan Luis Blanco]

Hi everyone,

today I completed all the glyphs in Adobe Latin 3 character set, made small tweaks and enhancements in several glyphs and revised and refined the spacing. I'll start tomorrow with the kerning. Please find attached a PDF with the whole character set and three pages of text in six different languages.

File is here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v022.glyphs

Best,

Juan

[vaishnavi Murthy]

Hello,

Akaya Github link has been modified to separate the Kannada and Telugu files.

Updated Links :

Kannada --- https://github.com/vaishnavimurthy/Akaya-Kannada

Telugu ------ https://github.com/vaishnavimurthy/Akaya-Telugu

Latin -------- https://github.com/Txonliz/Akaya-Latin-font

Thanks !

Best,

Vaishnavi

[梁海]

@Vaishnavi:

Just a quick reply. I hope some of them will be helpful.

But, basically, you need to really read Microsoft's Indic OTL documentations to understand how Indic shapers work.

And, for reference, although it's not my best piece of work, but Hind Guntur could still be helpful: https://github.com/itfoundry/hind-guntur

Murty Telugu's shaping is really good, but it's not open-sourced therefore you might need to dig into its OTF file using tools like FDK's "spot".

<<AI composite glyph -- not replacing>>

— Your "lookup abvs_8;" (https://github.com/vaishnavimurthy/Akaya-Telugu/blob/master/Source/Features/Akaya_Telugu_GSUB.txt) is wrong. tlmE is already ligated to consonant bases in your "lookup abvs_7".

<<Halanth breaking a er vottu. Similar issue in Kannada>>

— Of course. Your "lookupflag IgnoreMarks;" is not on. You need to either ignore vottu marks or reorder the halanth to the left side of all vottue marks.

<<Composite CV anchors -- not recognising>>

— Your "GPOS" table has only "DFLT" script registered.

<<How to place the AI length mark below the vottu? >>

— Ligating, reordering, positioning, use whichever method you want. I do reordering.

<<How to place the TAIlengthmark that supposed to appear here? >>

(TAc2_AIlengthmark?)

— Looks like the same issue to me.

<<How to place the TAIlengthmark that supposed to appear here? >>

(<AIlengthmark, RAc2>?)

— Since your AIlengthmark is already supposed to be ligated to the base ("consonant_AI" ligatures), you should anchor RAc2 to "consonant_AI" ligatures.

<<All the KA forms need to have the alternate SSA mark as de ned currently in the AKHN feature. How to make this work? >>

— Your "lookup akhn_ssa_alt" and "lookup akhn_ssma_alt" are all wrong. Glyphs like SSAc2 haven't even formed yet in the stage of the "akhn" feature. Basically these two lookups should be executed after all of your CV ligating rules (lookup abvs_*). Read Telugu's OTL doc please.

--

You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/a5c3a4c9-5901-4d9c-a923-f8b744973085%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[梁海]

@Vaishnavi: Btw, I feel you have some misunderstanding of GSUB mechanism (for example that "lookup abvs_8"). Please read http://opentypecookbook.com thoroughly.

--

梁海 Liang Hai

[vaishnavi Murthy]

Hi Liang,

Thank you for going through the files and the very useful suggestions. I'll try to follow them.

I'm learning how to write to feature files as I'm working on Akaya. I did go through the links you've suggested earlier. I see that there are several way to get the same thing to work and it gets confusing for me here. Also, the order of putting things gets tricky therefore. Like the precomposed mark ligatures doesn't work unless I place it much before. Thank you for pointing-out 'lookup abvs_8', I've placed if before forming the C+vmE. Now to get the rest to work :)

Will post updates soon.

Best,

Vaishnavi

[梁海]

@Vaishnavi:

"Like the precomposed mark ligatures doesn't work unless I place it much before" — uh, not sure what you meant here…

About "lookup abvs_8", though, I should say, you really don't need to split those rules into "lookup abvs_1 … abvs_11". Just use a single lookup and it'll do the job.

--

You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/4845a7bc-f2c3-485d-ab4c-150c9e7f9d2c%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[vaishnavi Murthy]

On Tuesday, 22 December 2015 10:36:22 UTC+5:30, Liang Hai wrote:

@Vaishnavi:

"Like the precomposed mark ligatures doesn't work unless I place it much before" — uh, not sure what you meant here…  

About "lookup abvs_8", though, I should say, you really don't need to split those rules into "lookup abvs_1 … abvs_11". Just use a single lookup and it'll do the job.

True. I split them up to make it easier to test and reorder if required. I'll combine them once I have a working flow. Thank you for pointing this out.

 

On Mon, Dec 21, 2015 at 8:13 PM vaishnavi Murthy <vaishna...@gmail.com> wrote:

Hi Liang,

Thank you for going through the files and the very useful suggestions. I'll try to follow them.

I'm learning how to write to feature files as I'm working on Akaya. I did go through the links you've suggested earlier. I see that there are several way to get the same thing to work and it gets confusing for me here. Also, the order of putting things gets tricky therefore. Like the precomposed mark ligatures doesn't work unless I place it much before. Thank you for pointing-out 'lookup abvs_8', I've placed if before forming the C+vmE. Now to get the rest to work :)

Will post updates soon.

Best,

Vaishnavi

--
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.

To unsubscribe from this group and all its topics, send an email to googlefonts-discuss+unsub...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.
To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/4845a7bc-f2c3-485d-ab4c-150c9e7f9d2c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[Juan Luis Blanco]

Hi everyone,

the final version of Akaya Latin is here. There have been a couple of last minute changes: new designs of ‘M’ and ‘g’ to match better the overall style of the typeface.

The kerning is also completed. Hinting decisions will be made when Latin is put together with Kannada and Telugu. In the PDF attached you'll find samples of mulitlingual texts (English, German, Danish, Polish, Czech, Hungarian, French and Spanish)

Glyphs file here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v025.glyphs

Merry christmas to everyone!

Juan

[vaishnavi Murthy]

Wishing everyone a very merry happy happy Christmas :)

Akaya Telugu + Latin is up for initial round of testing : https://github.com/vaishnavimurthy/Akaya-Telugu

A few issues are already marked here : https://github.com/vaishnavimurthy/Akaya-Telugu/blob/master/Tests/Akaya_Telugu_20151125.pdf

Thanks!

Best,

Vaishnavi

[梁海]

@Vaishnavi:

Btw, it's said "Not included … TELUGU SIGN CANDRABINDU" on page 5, however "U+0C01 TELUGU SIGN CANDRABINDU" / ARDHANUSWARA / ARASONNA is already included in your font.

"ISSUE : Halanth breaking after vottu. Similar issue in Kannada" — I suppose you're already working on it? There shouldn't be any technical obstacle for you though.

--
You received this message because you are subscribed to a topic in the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/googlefonts-discuss/8GhUoeMZy-0/unsubscribe.

To unsubscribe from this group and all its topics, send an email to googlefonts-dis...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/82e53e93-9216-4761-a9f8-fa44e9a87d61%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[Vaishnavi Murthy]

Hi Liang,

The character mentioned there is U+0C00 and not U+0C01. This is a Vedic mark. You read more about this here : http://www.unicode.org/L2/L2010/10392r2-chandrabindus.pdf

For the Halanth breaking, I really don't have a solution. When I flag it with ignore marks, all the anchors stop working. I don't understand why this is.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAH7k%2BnePR6uE4Y7H3c2aRFuGep1XhaCUvNRYdpypJ27oChQQrA%40mail.gmail.com.

[梁海]

@Vaishnavi:

You've already mentioned "U+0C00 TELUGU SIGN COMBINING CANDRABINDU ABOVE" in the same line (the last line, page 5), and since you're apparently using Unicode character names, "TELUGU SIGN CANDRABINDU" here (in the last line) is a duplication to "U+0C01 TELUGU SIGN CANDRABINDU" / ARDHANUSWARA / ARASONNA mentioned in the top of the page 5.

I know the difference between the misnomer (because of ISCII) "U+0C01 TELUGU SIGN CANDRABINDU" ("ARDHANUSWARA/ARASONNA") and the newly introduced Vedic mark "U+0C00 TELUGU SIGN COMBINING CANDRABINDU ABOVE".

About "the Halanth breaking", did you make sure all C2 glyphs you want "sub consonant halant by consonant_halant;" to ignore *and* all mark glyphs that you want to anchor (using a "_"-starting anchor name) to base or mark glyphs are defined as GDEF class 3?

Btw, since I can't see your full workflow online, I'm not sure if your way of defining GDEF table is interfering your method of generating mark positioning rules.

You can provide a testing build (in a branch, for example) so people can see what really happened to make "all the anchors stop working".

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAO8-OAN9LCmv1%3Dfp5fp2PNevOra1%2BU904Rj3Tsrn54ROnnd1JA%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

梁海 Liang Hai

[Vaishnavi Murthy]

On 26 December 2015 at 11:06, 梁海 <lian...@gmail.com> wrote:

@Vaishnavi:

You've already mentioned "U+0C00 TELUGU SIGN COMBINING CANDRABINDU ABOVE" in the same line (the last line, page 5), and since you're apparently using Unicode character names, "TELUGU SIGN CANDRABINDU" here (in the last line) is a duplication to "U+0C01 TELUGU SIGN CANDRABINDU" / ARDHANUSWARA / ARASONNA mentioned in the top of the page 5.

I know the difference between the misnomer (because of ISCII) "U+0C01 TELUGU SIGN CANDRABINDU" ("ARDHANUSWARA/ARASONNA") and the newly introduced Vedic mark "U+0C00 TELUGU SIGN COMBINING CANDRABINDU ABOVE".

Aha. I get it now. 

Arasonna has two graphical variants as you must be knowing already. U+0C00 & U+0C01 are phonetically very different and have only recently been separated. Because of this, in a lot of places, the encoding confusion is already there. So, we have two phonetic values mapped on U+0C01 seen in a few documents with the two varying graphical representations. This causes confusion in the searchability and interpretation of a digitised document. For this, U+0C01 can support both the forms in a font of this kind to record this ambiguity.

To take this a step further, it'll be interesting to map the phonological variations of various languages over time/regions into a script/font. For example, in Kannada, ಱ = RRA (trill) is simply replaced with ರ = RA or a double RA i.e., ರರ್. Here, once again, there is no way for me to present it in its current graphical form while also capturing its phonetic value in most typefaces. 

An typeface/interface that'll allow for this might also be useful to document Non-native pronunciations maybe.

About "the Halanth breaking", did you make sure all C2 glyphs you want "sub consonant halant by consonant_halant;" to ignore *and* all mark glyphs that you want to anchor (using a "_"-starting anchor name) to base or mark glyphs are defined as GDEF class 3?

Btw, since I can't see your full workflow online, I'm not sure if your way of defining GDEF table is interfering your method of generating mark positioning rules.

You can provide a testing build (in a branch, for example) so people can see what really happened to make "all the anchors stop working".

I don't understand what GDEF class 3 is. Does it define all the spacing marks and non-spacing? I wanted to know if there's any resource that lists all the syntaxes.

I've uploaded 2 versions I was testing with varying @Marks table. Akaya_Telugu_1 and its corresponding TTF file have an extensive table and Akaya_Telugu_2 with fewer letters. Both of them have issues with breaking anchors.

 

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAH7k%2BneuXpLhK2yA5T_%2BS0ZvEJsfSQiFGseWOU2g-X-GaOy4ag%40mail.gmail.com.

[vaishnavi Murthy]

Forgot to add the link : https://github.com/vaishnavimurthy/Akaya-Telugu/tree/master/Tests/20151226_All%20feature%20files

(I've uploaded 2 versions I was testing with varying @Marks table. Akaya_Telugu_1 and its corresponding TTF file have an extensive table and Akaya_Telugu_2 with fewer letters. Both of them have issues with breaking anchors.)

[Juan Luis Blanco]

Hello,

Akaya Latin has a new ampersand. The yen symbol has been corrected and some missing kerning pairs added. Final version is here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v026.glyphs

Happy new year to everyone,

Juan

[vaishnavi Murthy]

Hello,

Akaya Kannada 1.0 & Akaya Telugu 1.0 are up for testing :

https://github.com/vaishnavimurthy/Akaya-Kannada

https://github.com/vaishnavimurthy/Akaya-Telugu

I haven't managed to get the virama after a vattu to combine with the base glyph.

Currently, the flags that are used here are making all the anchors to stop working.

Any solutions to this will be welcome. 

Dave, do let me know what to do next.

Thanks.

Best,

Vaishnavi

[Dave Crossland]

Hi

Awesome, thanks!

I need from you a few paragraphs of description of both families for use in your specimen page - for example https://www.google.com/fonts/specimen/Playfair+Display - and your G+ About page updated, and your twitter handle confirmed :) 

On 2 January 2016 at 21:21, vaishnavi Murthy <vaishna...@gmail.com> wrote:

Hello,

Akaya Kannada 1.0 & Akaya Telugu 1.0 are up for testing :

https://github.com/vaishnavimurthy/Akaya-Kannada

https://github.com/vaishnavimurthy/Akaya-Telugu

Dave, do let me know what to do next.

Thanks.

Best,

Vaishnavi

On Wednesday, 30 December 2015 20:53:55 UTC+5:30, Juan Luis Blanco wrote:

Hello,

Akaya Latin has a new ampersand. The yen symbol has been corrected and some missing kerning pairs added. Final version is here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v026.glyphs

Happy new year to everyone,

Juan

--
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 https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/cb0912b8-60ba-424b-9b14-80029818d6b2%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Erin McLaughlin]

Hi Vaishnavi!

Dave has asked me to take a look at some of the Indic projects that are being finalized. I have a few suggestions for the Akaya families that you may (or may not!) want to consider implementing :)

Thanks so much for sharing your project, it's beautiful!!!

– Erin

On Monday, January 4, 2016 at 11:18:22 AM UTC-6, Dave Crossland wrote:

Hi

Awesome, thanks!

I need from you a few paragraphs of description of both families for use in your specimen page - for example https://www.google.com/fonts/specimen/Playfair+Display - and your G+ About page updated, and your twitter handle confirmed :) 

On 2 January 2016 at 21:21, vaishnavi Murthy <vaishna...@gmail.com> wrote:

Hello,

Akaya Kannada 1.0 & Akaya Telugu 1.0 are up for testing :

https://github.com/vaishnavimurthy/Akaya-Kannada

https://github.com/vaishnavimurthy/Akaya-Telugu

Dave, do let me know what to do next.

Thanks.

Best,

Vaishnavi

On Wednesday, 30 December 2015 20:53:55 UTC+5:30, Juan Luis Blanco wrote:

Hello,

Akaya Latin has a new ampersand. The yen symbol has been corrected and some missing kerning pairs added. Final version is here: https://github.com/Txonliz/Akaya-Latin-font/blob/master/Akaya_L_v026.glyphs

Happy new year to everyone,

Juan

--
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-discuss+unsub...@googlegroups.com.

To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at https://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/cb0912b8-60ba-424b-9b14-80029818d6b2%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Cheers
Dave

[Vaishnavi Murthy]

Wow! I didn't notice the NYA! Thanks :D

The others, when it comes to strokes angles, its just a convention I follow here. I'd like to retain these after having tested several variations.

This style of Telugu has been a nightmare to fix. Overall, it seems to work ok for now. I know about the collisions/kerning issues & I'm trying find a good solution for this. 

Will have these updated along with the two names soon.

Thanks again Erin. Its always great to get some good feedback.

Best,

Vaishnavi