Google Font updates: Exo

[Robin Mientjes]

Hi all,

I’m Robin. I run the Tiny Type Co. and have been making type for about nine years.

Today I’ve started working on updating Exo. Dave requested I mostly focus on spacing and proportion, so that’s where I’m beginning.

But the real first task is to convert some old FontForge files to UFO files. This will make it easier to organise the project.

When I’m done with this, I expect to create – for now – two two-master design spaces in Superpolator, and match the instances closely to the old intermediate weights. This will require me to make sure the masters are compatible. I’ll add to this post later today on my progress.

[Robin Mientjes]

Dave has asked to not change the design (too much). Some of this is unavoidable if we want clean, compatible masters. The existing Black is too narrow and too tightly spaced, and a lot of attention has to go into that style to make the intermediates spaced better. If it ends up feeling too wide in the extremes, we can set a Regular master that’s intermediate, and the extremes can be spaced more tightly. But I’m not there yet!

I’m figuring out the right proportions, optical corrections, overshoots and diagonals for now. The only design choice I’m making that might be noticeable is the /g/. If anyone has a better idea than this change (attached before and after). Am I changing too much?

Thin /g/ before

Thin /g/ after

Black /g/ before

Black /g/ after

[Robin Mientjes]

It’s worth noting I’m keeping pretty granular track of my changes (for now) on Github → https://github.com/rbmntjs/Exo-2016

[Alexei Vanyashin]

Black /g/ after

I think the new /g is a big step forward. To me this change isn't too much. 

I would give more compensation to the horizontal strokes of the bottom loop of /g, similar to what is done with /a, /e.

[Jacques Le Bailly]

Hi Robin,

welcome to the project ! It is already looking good !

// Jacques

[Dave Crossland]

Can we keep the old g as a stylistic alternative? 😆

[Robin Mientjes]

On Friday, November 18, 2016 at 8:15:56 PM UTC+1, Alexei Vanyashin wrote:

I would give more compensation to the horizontal strokes of the bottom loop of /g, similar to what is done with /a, /e.

Thanks, I’ll note it down. There are a few glyphs where I think I have to apply the same logic, such as /s and /S when compared to /e, /c, /C.

Thanks, Jacques! Glad to be here. And Dave, well, sure, but I’ll have to find a way to 1: keep it balanced with /a, /s, /e, and 2: keep it feeling similar all across the weight range, which the old /g doesn’t really do. Suggestions are definitely welcome.

[Robin Mientjes]

Over the weekend I worked some more on general spacing matters and optically correcting the Black weight. Today I’ll be continuing down the path of making the Black look like a plausible weight increase from the Light. Still alphabetic work, but after the base is done I’ll invite more critique and focus on the diacritics and numbers.

[Alexei Vanyashin]

Thanks, Jacques! Glad to be here. And Dave, well, sure, but I’ll have to find a way to 1: keep it balanced with /a, /s, /e, and 2: keep it feeling similar all across the weight range, which the old /g doesn’t really do. Suggestions are definitely welcome.

By looking at Exo 2.0 as an evolution of 1.0 the legacy /g looks rudimentary. Best is too ask the original author. 

[Dave Crossland]

I agree. Natanael? :) 

--
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 googlefonts-discuss@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/CADUrt2GExahoieW8BJ_XaffAQzsDDOXJn78BZyU_2r2FGpY1PA%40mail.gmail.com.

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

--

Cheers
Dave

[Natanael Gama]

Hey Guys!

As the author of the exo font I think the design essence of the font, thought not that well executed, is that it is based on the idea of a square with round corners. It can't be too much humanistic, it has to preserve this sort of sturdy geometry (which I got a bit rid off in Exo 2).

So my opinion is that it is getting to rounded and loosing the feeling people seem to like once again.

I would like to see flat lines in the curves, or curves that appear that way, as it was in the original design.

For the rest it looks like it is going in the right track!

[Natanael Gama]

I am not saying that what was done was wrong in anyway :) It's just to resist the temptation of making it too much rounded

[Dave Crossland]

Yeah I think keeping Exo 2 in mind and staying more to the Exo 1 design and away from the Exo 2 design is crucial to this project

Its interested to see how Exo 1 compares to it in https://fonts.google.com/analytics is instructive - they are nearly the same and Exo 1 is actually more popular 

[Robin Mientjes]

I’m listening, Natanael. It’s your design, I’m trying to interpret it best I can. So please keep voicing your intentions, and I’ll do my best, with feedback from the community, to draw it correctly. Here’s a /g with slightly more square-ish looks. Is this a better direction for you? (Ignore, for now, the weight difference between the bottom bowl and the connecting curve – I made a rash decision right before I took the screenshot and it didn’t pay off, but here’s the screenshot anyway.)

 

[Dave Crossland]

I wonder if the w/W should become a little more conervative (with the original as ss01) and the boxy 'g' should be kept as default with a more conservative 002...

Perhaps even a 'human/robot' axis of variation? :D

[Dave Crossland]

On 21 November 2016 at 12:33, Robin Mientjes <robmi...@gmail.com> wrote:

Here’s a /g with slightly more square-ish looks. Is this a better direction for you?

For me: Perfect!! :)

[Robin Mientjes]

Today I’ve added a lot of anchors and redrawn the diacritics. Most anchors are now in place, and I’ve started working out the base glyphs of the GF Latin Plus character set.

Not happy yet with the new cedilla, but I’ll get there. There are already a few stylistic sets in Exo that I’m going to avoid for now, because the required diacritics for some of them create an ungainly mess of composite glyphs, but when the general glyph design is fixed, it’ll be easy to add these options, and include alternate /w, /g, etc.

I’m also skipping all sups and subs, and the small caps. They can be generated from the compatible masters, and then it’s just some adjustments and more anchor work. Unless someone has a magical workflow for this, I’ll also be waiting with that for now.

[Alexei Vanyashin]

For what you mentioned just the usual RMX Tools.

Speaking of 'magical' tools, there is HT LetterSpacer, a UI script for auto-setting metrics in Glyphs App.

--
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/WBpnBnSr8Js/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 googlefonts-discuss@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/a8310fff-caed-4e5e-ab54-4d2a522a9a31%40googlegroups.com.

[Natanael Gama]

I think it's very good now :)

[Robin Mientjes]

Today’s work has been a bit more conservative: the numbers are a very particular design and I’m trying to not disturb the design too much. So far I’ve mostly focused on the default lining figures. I’m not sure how the old-style works, because it’s closer to 3/4s and, well, it’s only available through OpenType.

I’ve also set up anchors on base glyphs, most combining diacritics† and the mathematical symbols.

† What rules do people normally follow for the Vietnamese set? All I see is the base diacritics: are those combined per glyph, or do you pre-combine the combined diacritics, keeping another set of zero-width glyphs (in three cases)?

[Alexei Vanyashin]

We do pre-combine stacked Vietnamese accents:

For more specifics see Nhung's deck.

Here is a full list of VIT diacritics we use, including pre-combined accents. 

gravecomb

acutecomb

circumflexcomb

brevecomb

tildecomb

hookabovecomb

horncomb

dotbelowcomb

strokeshortcomb

brevecomb_acutecomb

brevecomb_gravecomb

brevecomb_hookabovecomb

brevecomb_tildecomb

circumflexcomb_acutecomb

circumflexcomb_gravecomb

circumflexcomb_hookabovecomb

circumflexcomb_tildecomb

--

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/WBpnBnSr8Js/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/b5085de1-cb4d-4afc-8f94-6e2c77f12416%40googlegroups.com.

[Robin Mientjes]

Thanks, Alexei. That’s exactly what I needed. The deck is very useful too.

What I need to know next is from Natanael: I’m going through the numbers now, and while the default lining and tabular lining make sense to me, I am not sold on the old-style figures. I know we have to keep them, in order to maintain continuity with this update, but I don’t understand the rules for your offsets. How do you envision the 3/4 old-style set to work?

[Alexei Vanyashin]

Running ahead,

Here is a note on the GF Glyph Sets for non-GlyphsApp users.

GF Glyphs Sets(filter lists) follow Glyphs App naming scheme GlyphsData.XML and are built in such a way so that when you paste in the list, all the OT code will be built automatically by the app. For example 'zero.zero' in the list filters represents a slashed zero, and triggers relevant auto-feature generation in Glyphs App. 

'hyphentwo' is the Glyphs App name for standard hyphen(U+2010). In Glyphs the name 'hyphen' is assigned to U+002D, instead of U+2010(as suggested by unicode). Since U+2010 is required by CLDR both hyphen and hyphentwo are included in the GF Glyph Sets. 

Here is the recommended naming for these these duplicate hyphens:

hyphen or uni002D

hyphentwo or uni2010

Should you have any questions on the glyphs sets, checkout their unicode values in the .NAM files.

[Robin Mientjes]

I started coughing Friday afternoon and it finally started letting up yesterday morning. I’ve spent a lot of time since last week on the italic masters. They require the same adjustments – taller x-height on the heavier weights, curve corrections to stay continuous, and evened out thin and thick contrast. I’ll attach a showing later today, but I’ve been pushing changes on the regular to the git repository if anyone is curious to see the process like that.

[Dave Crossland]

Thanks Robin! Full speed ahead :)

[Robin Mientjes]

There’s a lot of work that goes into the italic poles. I decided to tackle these separately as there are quite a few incompatible glyph designs and I think we should draw these with the extrema in place – not just slant and fix later. It makes it more predictable for interpolation, in my experience, as I can match the off-curve proportions between the poles. I might still slant and fix some upright designs, such as the numbers, diacritics and currency, but the core alphabet is getting its own cleanup.

I spaced some things a bit too tight, which this GIF revealed to me. So those are already being tightened back. But yeah, this is trucking on steadily. I hope to have the base alphabets and numbers for all four poles ready during next week for a good critique round, maybe with a regular and bold interpolated from them. Then we can judge the overall spacing, in particular, as that is the thing that has changed the most, for existing users.

[Robin Mientjes]

After some gritting of teeth and some duplicate work, I’ve now brought the base alphabets of the four poles close to proposal. I’m going to take a lot of the upright glyphs into the italic files, but I’d rather have sign-off before I do. I’ve attached a preview of the two design spaces – which means this can also trivially be generated as a variable font, whenever that comes around.

Any special requests while I prepare a demo PDF of the four poles + regular and bold? Seems like those are good references to judge the entire spectrum by.

[Natanael Gama]

Where are the editable files so I can make a few suggestions?
The UFOs on github doesn't open in glyphs
Thanks

[Robin Mientjes]

On Tuesday, December 6, 2016 at 3:08:40 PM UTC+1, Natanael Gama wrote:

Where are the editable files so I can make a few suggestions?
The UFOs on github doesn't open in glyphs

Could you start with the PDFs I’m attaching here? It’s your 1.0 and my 2016 update. For the four poles for now. I don’t know why the UFOs don’t open, and am not in a position to research.

 

[Dave Crossland]

On 6 December 2016 at 09:39, Robin Mientjes <robmi...@gmail.com> wrote:

Could you start with the PDFs I’m attaching here? It’s your 1.0 and my 2016 update. For the four poles for now. I don’t know why the UFOs don’t open, and am not in a position to research.

Perhaps try a demo version of RoboFont to open them?

(Rob, any special extensions needed to open them beyond that...?)

[Robin Mientjes]

On Tuesday, December 6, 2016 at 8:00:59 PM UTC+1, Dave Crossland wrote:

any special extensions needed to open them?

Nothing from my side. The UFOs are vanilla, the only thing that prevents them from building is that the features don’t match up with the glyphset – a result of also correcting the family to the latest GF Latin set. Maybe Glyphs chokes on that?

[Dave Crossland]

Could be - Natanael, try deleting / blanking the features.fea file and reloading? 

[Robin Mientjes]

The process the past few days has involved tweaking the numbers more, to get a cleaner interpolation and more squared shoulders, and to try to cover as many of the composables as possible. What’s left is decomposing some of the stranger ones (Lslash, lslash for example into L, l and combining strokes) and adding the Vietnamese and further GF Latin Plus combining diacritics.

[Natanael Gama]

Hi Rob,

I think some letters should have more open straight counters, like exo 2. there is a slight roundness on the upper part of the 'a' that i don't like. i also happens on the 'c', maybe others. I now it is a strange shape, the inside straight and the outside rounded, but it is possible do design with better quality than I did.

I think you could be more brave on the correction of the proportions, specially of the black weight.  W and w look too dark, along with many other shapes like M and R (but there are more). Please do make optical corrections in terms of weight otherwise the rendered text won't improve much besides the better spacing and kerning. 

Keep up the good work.

[Robin Mientjes]

On Monday, December 12, 2016 at 12:55:37 PM UTC+1, Natanael Gama wrote:

I think some letters should have more open straight counters, like exo 2. there is a slight roundness on the upper part of the 'a' that i don't like. i also happens on the 'c', maybe others. I now it is a strange shape, the inside straight and the outside rounded, but it is possible do design with better quality than I did.

I added those straight sides now, let’s see how that works. Need to tweak the proportions a bit more, but your logic is back.

 

I think you could be more brave on the correction of the proportions, specially of the black weight.  W and w look too dark, along with many other shapes like M and R (but there are more). Please do make optical corrections in terms of weight otherwise the rendered text won't improve much besides the better spacing and kerning. 

I’ve been spending a lot of time on that yesterday, actually. It takes some tweaking, but I’ll try to make a sampler later. Worry not about the rest of my efforts: the text will improve quite a bit in texts without M, W and R as well.

[Robin Mientjes]

I’ve been spending a lot of time on that yesterday, actually. It takes some tweaking, but I’ll try to make a sampler later. Worry not about the rest of my efforts: the text will improve quite a bit in texts without M, W and R as well.

In fact, here’s a quick GIF showing the changes. There are still some dark spots, but it’s going places. 

[Robin Mientjes]

I’m getting to the point where we need to start deciding on a few things.

First of all, am I straying too far from the original? I’ve had to modify a lot of things in order for the spacing to be changed consistently.

Secondly, am I to retain all the alts, idiosyncrasies and bugs from the original because of legacy reasons? I can give you a whole laundry list of issues that will take time and deliberation but I’m curious to hear ideas about this first. Before I commit to replicating.

And last, I’ll be tweaking the regular weight, turning that into a proper pole and interpolating from that for a 3-pole range for the two styles, so that I can match the spacing more closely to the original, but before that happens I do think it’s sensible to settle on the extremes, as much as possible anyway. But that’s my usual workflow, so I’m wondering if there are any tricks that people recommend. For example, I have my base glyphs set up with anchors, and the 55 required diacritics from the GF Latin Plus set, but I don’t have any clever stuff to auto-build the composites.

[Natanael Gama]

Design wise it is much better, thanks. some more tweaks here and there and you're done. Notice the lowercase 'e' (black master) could have a bit more of white space inside, a bit more like the lowercase 'a' rotated (still a totally different shape, only in terms of black/white balance) :) 
Also on the black master, make the 'S' and 's' terminations like the 'a' and the 'c'. Also now they are not coherent, lowercase s is straight and the uppercase is a little rounded. Preserve the original design on this, only optically corrected.

Personally I would abandon all the alternates, I am not sure what Dave thinks about it. Exo started as a way to discover typography and I did a lot of things out of a need to know and understand how things worked. I would stick to the default shapes only.

Feel free to put here all your decisions and I will comment on them.

About your technical concern, I just use make component glyph in Glyphs app and it automatically creates all the diacritical characters. Was that your question?

Cheers

[Natanael Gama]

Diagonals still look overall too bold. I would reccommend to print a full whole text using all the letters (you can use adhesion text) on like 12 points and mark all the dark spots, then fix it on screen. You can do the same on a screen but because it is more tricky i would start on a piece of paper.

[Robin Mientjes]

On Tuesday, December 13, 2016 at 3:36:59 PM UTC+1, Natanael Gama wrote:

Design wise it is much better, thanks. some more tweaks here and there and you're done. Notice the lowercase 'e' (black master) could have a bit more of white space inside, a bit more like the lowercase 'a' rotated (still a totally different shape, only in terms of black/white balance) :) 
Also on the black master, make the 'S' and 's' terminations like the 'a' and the 'c'. Also now they are not coherent, lowercase s is straight and the uppercase is a little rounded. Preserve the original design on this, only optically corrected.

Yep, still going through a lot of revisions for the diagonals. Some of the choices interpolated better but made for a less even Black pole, so I might have to split the difference for a few glyphs by inserting a Regular pole just for them.

 

Personally I would abandon all the alternates, I am not sure what Dave thinks about it. Exo started as a way to discover typography and I did a lot of things out of a need to know and understand how things worked. I would stick to the default shapes only.

This is what I hope to do, but out of the 100M page views a week, how many use the OpenType alts? That’s a genuine concern. I don’t know if we have numbers on this.

 

About your technical concern, I just use make component glyph in Glyphs app and it automatically creates all the diacritical characters. Was that your question?

 

I use RoboFont and the only thing I wish was a bit more automatic is the component glyph logic – may just be a fun Christmas programming challenge.

I’ll update this soon with more samples of the diagonals, and the new glyphs in the GF Latin Plus set.

[Robin Mientjes]

I’ve been having a heck of a time with the diagonals. In the process, I’ve also tweaked a lot of other interpolating forms, and I’ve implemented the list of design requests Natanael made earlier – a lot of letters are more squarish like the original. I’m also close to a complete GF Latin Plus set, barring the choice about figure forms and required alternates. After that, it’s kerning and fine-tuning the intermediate weights.

[Dave Crossland]

I don't think it's good  to abandon the alts, no. :)

I think the M in black is a little too heavy compared to everything else

On Dec 16, 2016 9:11 AM, "Robin Mientjes" <robmi...@gmail.com> wrote:

I’ve been having a ball of a time with the diagonals. In the process, I’ve also tweaked a lot of other interpolating forms, and I’ve implemented the list of design requests Natanael made earlier – a lot of letters are more squarish like the original. I’m also close to a complete GF Latin Plus set, barring the choice about figure forms and required alternates. After that, it’s kerning and fine-tuning the intermediate weights.

--
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 googlefonts-discuss@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/a117270c-69be-4efb-8400-567623705b62%40googlegroups.com.

[Robin Mientjes]

On Friday, December 16, 2016 at 12:30:52 PM UTC+1, Dave Crossland wrote:

I don't think it's good  to abandon the alts, no. :)

They’re adding a lot of weight to the work. I’m in this project to fix it, and no matter how easy that may have seemed from the outside, it required redraws on pretty much every glyph. I’ll be redoing the kerning as well, I have to still draw all the numeral variants and the small caps, which don’t reduce so well in the Black pole because it’s so dark. Those need kerning too, and working features. Then, when that is done, I have to match the interpolated instances as closely as possible to the optical weight of the current styles – at least the core weights.

Now I’m trying to redraw the spurless and the rounded glyphs and, well, they match with nothing. They’re not a simple matter of ‘fix the spacing’ – no matter how often that was described as my brief. The alts are a massive, inconsistent and inconsequential conceptual distraction, and now they are also a massive cost – for me.

Natanael, I’m asking again, before I get to work: what is the idea behind the ‘old-style’ numerals? They act a little bit like 3/4 numerals. Is that correct? And if so, what is the reference point for me to match them to optically? The descenders on Exo are pretty generous – almost as tall as the ascenders – which means the usual 3/4 proportion feels a bit floaty, if you know what I mean.

[Robin Mientjes]

Is there a recommended or default way for handling .sups, .subs, superior and inferior glyphs (four different vertical positions, and different naming schemes)? I’m going to generate them from .tnum but I don’t want to update four different sets of the same thing.

[Alexei Vanyashin]

See this section on .sups. The described workflow is for Glyphs App setting, so do ask questions if this differs from your setup.

On Tue, Dec 20, 2016 at 3:57 PM Robin Mientjes <robmi...@gmail.com> wrote:

Is there a recommended or default way for handling .sups, .subs, superior and inferior glyphs (four different vertical positions, and different naming schemes)? I’m going to generate them from .tnum but I don’t want to update four different sets of the same thing.

--

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/WBpnBnSr8Js/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/804da2ca-989d-4506-ac05-39f116b25237%40googlegroups.com.

[Jacques Le Bailly]

Is there a recommended or default way for handling .sups, .subs, superior and inferior glyphs (four different vertical positions, and different naming schemes)? I’m going to generate them from .tnum but I don’t want to update four different sets of the same thing.

If you use Glyphs, you can put them into the .dnom position. For the .numr you can choose to use outlines or components. For the ...superior and ...inferior you can use components. This is also easier for TT hinting afterwards. 

Unless you want typographically correct differences. Some people choose to have different sizes for superior/inferior and numerators.

[Robin Mientjes]

Thanks, Jacques and Alexei. Now I have GF Latin Plus + a few legacy things all finished for the upright poles. Italic poles will follow in close step, but they’re mostly derivative. I’ve set up and uploaded some glyph generation documents (Adieresis=A+dieresiscomb.case@top and so forth) for all the GF Latin Plus glyphs and their small caps. What’s a quick-and-dirty way to show my work so far? I’d like to generate a PDF without having to compose a specific one for this specific family (my own needs are different than what this family does).

[Alexei Vanyashin]

On Thu, Dec 22, 2016 at 2:22 PM Robin Mientjes <robmi...@gmail.com> wrote:

Thanks, Jacques and Alexei. Now I have GF Latin Plus + a few legacy things all finished for the upright poles. Italic poles will follow in close step, but they’re mostly derivative. I’ve set up and uploaded some glyph generation documents (Adieresis=A+dieresiscomb.case@top and so forth)

Glyphs App does this automatically. We have a few other quick generate commands listed in this tutorial

 

for all the GF Latin Plus glyphs and their small caps. What’s a quick-and-dirty way to show my work so far? I’d like to generate a PDF without having to compose a specific one for this specific family (my own needs are different than what this family does).

Drop your TTFs or OTFs http://www.impallari.com/testing/index.php in Latin-1 > Latin tab and make a screenshot.

[Dave Crossland]

You can also run notodiff on the original TTF and your latest build to show the differences :) please check in the files to the repo and link them here :)

--
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 googlefonts-discuss@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/CADUrt2H46jvNN%2BCAvN151GQs8jW2eRVfA5SJ%3D1G-GNi9AB3Q1A%40mail.gmail.com.

[Natanael Gama]

I think you should reduce the ascender and descender (maybe 20 Units), they look too tall. And then do the oldstyle

[Robin Mientjes]

On Friday, December 30, 2016 at 8:43:50 PM UTC+1, Natanael Gama wrote:

I think you should reduce the ascender and descender (maybe 20 Units), they look too tall.

I agree, I’ll go through this and even it out a bit.

 

And then do the oldstyle

 

But again: your design has them as 3/4-style, not oldstyle. I feel that the originals are not grounded enough – their ascenders and descenders are in-between in inconsistent ways. So, what do you think about that?

[Robin Mientjes]

Meanwhile, I’ve been setting up kerning groups today, fixing last anchor positions, and making a script to generate the OpenType features that involve suffxes. I’m still waiting for an answer on what to do with the not-really-old-style figures. Other than that, I’m much happier with the overal weight distribution in the Black pole and how it interpolates down. I’ll share some PDFs when Adobe is ready updating my apps…

[Robin Mientjes]

On Tuesday, January 3, 2017 at 3:42:24 PM UTC+1, Robin Mientjes wrote:

[…] a script to generate the OpenType features that involve suffxes

I’ll share my logic. It’s a very quick-and-dirty solution, but it works backwards based on the actual glyphs available, not a specific glyph set. That way, at least the targets are correct. You’ll still need to make sure the non-suffixed version exists for it to work.

https://gist.github.com/rbmntjs/bbd27a005e5c3e446103e2bf645d7162 

[Dave Crossland]

Hi Robin

Awesome! Please can you drop this in your forks repo in a scripts directory with a copy of the apache 2 license? 

--

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 googlefonts-discuss@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/4dbfe897-989a-4919-a8d5-be43f5374f49%40googlegroups.com.

[Robin Mientjes]

On Wednesday, January 4, 2017 at 11:40:57 PM UTC+1, Dave Crossland wrote:

Awesome! Please can you drop this in your forks repo in a scripts directory with a copy of the apache 2 license? 

 

Added it! It’s quick and dirty, and I’m pretty sure it does what Glyphs does automatically, but I like knowing what happens.

The interpolated exports are starting to look good. I’m doing tests at title level and text level, and my work has a side effect I hadn’t really focused on. Let me show you.

Except for the /w, most people will not notice the change. The overall reflow is limited, especially in heading settings. The only change is I think my interpolated Regular can be a tiny bit heavier. 

The real success is in text settings, where the original is frankly unusable.

So the respacing and new proportions work very well at text sizes down to 16. Below that I’m sure there are plenty of issues, but this is a great bonus. So in large use, the changes are mostly subtle and, for most people, invisible. For text use, it’s an entirely new font.

I’m going to upload PDFs later today, and then we can talk about final changes to the design.

[Robin Mientjes]

I’ve attached PDF samplers for the two upright poles, and included the small caps and figure styles. I’ll do a rough implementation of the .salts as well, but right now I want to focus on the overall design. So, Natanael, please red-pen this thing.

[Robin Mientjes]

While waiting for design feedback, I’ve been setting up thorough kern groups and going through my generation scripts to cull bugs. Several OpenType features are now more robust, but I’m not going to implement all of those just yet – some of it is more closely linked to the mastering, so I’ll discuss that next week. I’m also slowly tweaking the interpolated weights to match the originals – the Regular should now match, and I’ll measure stems soon for the other weights.

[Natanael Gama]

I like it very much, it keeps the original flair but in a much more consistent way, something that at the time I designed it I was clearly unable to do.
I don't remember what was my ideia with the oldstyle, but i think that with the ascender/descender corrected you can do standard old style figures (but keep the lining ones as default)

The small caps of the black weight need to be generally lighter (if you have remix tools just do a scaler on them)

Best

[Robin Mientjes]

On Friday, January 6, 2017 at 8:35:34 PM UTC+1, Natanael Gama wrote:

I don't remember what was my ideia with the oldstyle, but i think that with the ascender/descender corrected you can do standard old style figures (but keep the lining ones as default)

 

This is working for me so far – after I went through with the adjustments, it clearly was the right way.

Here’s the Demibold, for example.

 

The small caps of the black weight need to be generally lighter (if you have remix tools just do a scaler on them) 

 

I’ve tried reducing the overall weight but it doesn’t help – just makes the Black smcp look too thin. So I’ve been manually adjusting them, and I’m almost there for a next pass. Also spending a lot of time preparing kern groups, basic OpenType features and prep for the automatic glyph generation. These are also in the repository, for easy future reference. When I’m happy with the upright, I’ll make the same adjustments to the italic.

[Robin Mientjes]

I’d love to hear from Marc Foley regarding mastering requirements. I am working in RoboFont with the proposed glyph set from the font tools repository, I’m at the kerning and feature-writing stages of the design. But these are ‘pretty’ names. I hope the mastering and build flow have some provision for this? I’m not sure how you normally do this. For Tiny Type, I simply work with Unicode names according to the AGL guidelines, so that I have no renaming problems later on, so I’ve not been in this position before.

[Robin Mientjes]

I spent Friday and today focusing on getting all the various improvements ported over from the upright to the italic. The repository reveals a lot of commits. The only missing logic is some small caps that should have been generated, and some cap variants (right now, it’s just the parentheses, but I think we should include brackets and braces as well). The kerning will be extremely similar between the styles, and I’ll likely only generate four sets of kerning, to interpolate across the 18 files.

[Dave Crossland]

Awesome!!:)

[Robin Mientjes]

I am currently finalising the OpenType features, and I think I’ll have to match the existing implementations of everything, including the salt feature (as opposed to setting the variations up as ss##). So that’s relatively simple. But what about the font info? There’s a lot of values that need to update. It’s a redrawn version, so a 2017 release. What about authorship, license info, etc?

[Marc Foley]

Hey Robin,

I'll do a quick inspection today on your repo to see the setup. Please continue as normal.

Regards,

Marc

[Marc Foley]

Hey Rob,

I had a quick look at your project and overall, I'm super impressed.

I quickly took two of the .ufo instances and created a glyphs file.

The only issue I found was the starting points of the 'z'

Your OT code compiles perfectly.

Press ahead sir. I'll be in touch soonish.

[Robin Mientjes]

On Wednesday, January 25, 2017 at 1:41:12 PM UTC+1, Marc Foley wrote:

The only issue I found was the starting points of the 'z'

I’ve inspected this and can’t find it in the two poles. I’ve updated the repo to reflect some changes (and some Prepolator work), so hopefully this should be better now. What I’ve pushed is my release candidate for the uppercase. If I get sign-off on that, I’ll apply the same to the italic, which is largely the same kerning as the upright, and which has the exact same OpenType features. So that would be quick work.

Your OT code compiles perfectly.

 

Great. I hope it still does, but it works locally. I’ve split up some lookups and added back the last legacy sets. Now smcp doesn’t trigger parenleft.smcp, but c2sc does. Stuff like that. Pragmatic fixes. Keep me posted!

[Marc Foley]

I’ve inspected this and can’t find it in the two poles. 

 

this may have been caused by Glyphsapp import. I can fixes these small issues myself. 

I'm done with my inspection. Everything is awesome.

[Robin Mientjes]

Today I’ve brought the features and kerning in line with recent additions and fixes to the glyph set. Now it’s GF Latin Pro, plus all the legacy .salt glyphs. I am making sure it all interpolates cleanly, and so far it’s looking good. So Marc, it would be great if you could verify that the upright design imports and works properly. Then I can take my gathered knowledge and complete the italics as well – that’s only a hair away from done.

[Robin Mientjes]

After a few days of hard work to finalise the files, the Exo update is now ready for final reviews. Any export, interpolation, feature or kern issues can be addressed. Glyph set and glyph design, probably less so in this loop. Have a look and please let me know. There are TTF exports for easy preview.

→ https://github.com/rbmntjs/Exo-2016

[Robin Mientjes]

Some quick notes: font metadata is incomplete (I have no idea what it should say); features are mostly tested but, as usual, might come out different in different software; the GDEF table is giving me export grief, but only in the generated instances, not the masters, so I’m not sure if I’m missing something there (I know Glyphs has its own generation for that also, so maybe that’s where that’s handled?); the files are not style linked, but I’m not sure that that is a focus for these web fonts.

[Dave Crossland]

On 14 February 2017 at 11:28, Robin Mientjes <robmi...@gmail.com> wrote:

Some quick notes: font metadata is incomplete (I have no idea what it should say); features are mostly tested but, as usual, might come out different in different software; the GDEF table is giving me export grief, but only in the generated instances, not the masters, so I’m not sure if I’m missing something there (I know Glyphs has its own generation for that also, so maybe that’s where that’s handled?); the files are not style linked, but I’m not sure that that is a focus for these web fonts.

Marc can advise but https://github.com/googlefonts/gf-docs/blob/master/ProjectChecklist.md and http://fontbakery.appspot.com/ should get you started :) 

[Robin Mientjes]

I’ve spent all day now mastering the files. I’m slowly chipping away at the issues, but FontBakery thankfully reports verbose issues. There’s more work to do, and in particular I am struggling to match the old vertical metrics. Hope to figure out the right values soon.

[Dave Crossland]

On 15 February 2017 at 14:44, Robin Mientjes <robmi...@gmail.com> wrote:

I’ve spent all day now mastering the files. I’m slowly chipping away at the issues, but FontBakery thankfully reports verbose issues. There’s more work to do, and in particular I am struggling to match the old vertical metrics. Hope to figure out the right values soon.

Likely easiest thing is to open the UFO src in Glyphs, run Marc's glyphs script, export UFO, and see what it did :) 

[Robin Mientjes]

Today I am spending trying to match the existing fonts first in general terms – vertical metrics values, shared name table values – and then, when I’m happy, to write a set of name table things for makeOTF (or if I can right into the UFOs) to match weight values and weight names. Four poles, 18 files total, it makes for a bit of a puzzle.

[Robin Mientjes]

Already I am in a bind. If I have to match the vertical metrics of the original files, I will be forced to trim the caps that have double-stacked diacritics. FontBakery will then give me an error. So it’s either a full fail for the technical part (which I agree with), or an aesthetic problem of line increment reflow (which, well, that also sucks). What am I supposed to do in this?

[Marc Foley]

The OS/2 table has an fselection bit called "USE_TYPO_METRICS', https://www.microsoft.com/typography/otspec/os2.htm#fss

By enabling this bit, you can change the win asc and win desc to the bbox, You can then set the typo metrics to match the previous win asc and win desc values.

[Marc Foley]

Hey Rob,

When you're done, I'll port your .ufos to .glyphs. I think this may be the easiest solution for all of us. Thank you for all your hard work on this.

[Robin Mientjes]

To keep this in the public: I went back and forth with Marc in private e-mails, and we’ve now merged a Glyphs project into the repository. Please check it out! Marc’s quality assurance projects have been a huge help on this – now we match vertical metrics, glyph set (except for one erroneous glyph, which has been replaced with its correct implementation), and we’ve expanded the language support to GF Latin Pro, updated the horizontal proportions starting from the medium range and up, redrawn almost every curve, and respaced and re-kerned the entire family. It now works down to 16px, but looks and feels just like its old self at 32px and up.

Again, check it out. Let us know how this works. It’s been a long project, but it turns out that updating 4 masters and rebuilding their structure and redrawing and respacing, well, that’s a lot of work.