Greek Additions

[frankrolf]

This issue will track updates to the Greek Roman and the development of the Greek Italic styles. It is useful for this discussion to happen in public, so I expect this to be a longer-term issue, and be updated periodically with design PDFs, feedback, comments etc.

[frankrolf]

At ATypI Antwerp, I had a chance to meet with @irenevl and @thynem, where they suggested minor corrections to the existing Regular master:

These suggestions by Irene illustrate proposed changes to the capitals. The most significant change is happening in the construction of Omega, to avoid the balloon shape for the O-part. I am fine with the change in construction, but find the horizontals a little thin now. Most other adjustments are about letter proportion, and I fully understand why they should happen.

---

Irene also suggested changes to the lowercase. Especially the delta is much more balanced in my opinion. Other changes are structural (the alpha is now more upright), and proportional (upsilon, phi, psi).

---

Emilios may have found a good solution for the terminals of zeta, xi, and final sigma. Both Irene and Emilios remarked that the terminals I had drawn looked artificial – indeed their flag shape does not relate to any other element of Source Serif.

---

I hope this review format is useful, I am looking forward to your changes appearing in the greek-italic branch, we can continue our conversation at that point.

[frankrolf]

I have found some time to analyze the suggested changes to the Regular master in detail. Here is a comparison: (please note this comparison is extremely simplified and uses no kerning etc.)

• Irene changed a number of lowercase letters for better proportions (most visible in ευφψ. I am in favor of all these changes.
• The δ suggested by Irene is clearly superior.
• The asymmetric ω strikes me as a bit dark

Additionally, Emilios made some alternate suggestions for ζξωςϛω – this is mostly about the flags below the baseline. It is true that the flags I drew don’t have much of an echo in other parts of the alphabet – I was thinking of them being most related to πστ. I trust your decision on what those flags should be – obviously you are the experts.

• for the overall shape of the ζ, I prefer Irene’s solution
• for the overall shape of the ω, I prefer Emilios’.
• the change in the β seems minor to me, but let’s do it if it’s needed for clarity.
• the change in the α also seems minor. I feel making the counter all smooth goes a bit against the overall construction of the typeface. The new verticality of the letter probably would require some re-spacing.

[frankrolf]

This animation may be a bit hard to follow, here is a PDF: 2018-10-17 source serif greek comparison.pdf

[davelab6]

Thanks Frank! I love the animation :D

[thynem]

First version of Source Serif Greek Italic, for the Regular weight, is now on github.

Irene and I have submitted two approaches over the same design direction, with some alterations in order to open a discussion and find the optimal solutions going forward. Comparison is shown here:

For a more detailed preview, character sets and running text, you can download the PDF: SourceSerifItalicsGreekTestDoc181026.pdf

Some general comments for some of the glyphs over my approach:

• α, γ + ν and κ follow in some respect design elements a, v and k. ω borrows stroke ending idea from y.
• For ζ and ξ I tried to incorporate characteristic elements from z. For ζ it was fairly easy, however for ξ, due to the lack of space, this solution was not viable; a different direction was followed. I do prefer when ζ and ξ share design elements, so on these I would go with Irene’s solution which is more harmonised.
• The closed φ is not such a common solution – however for this design, it might be a suitable candidate for consideration.
• Baseline stroke endings follow the ones from the Latin while the descending ones make use of a more round approach: Frank had a μ math glyph in the original master with this approach, which I found interesting and thought could be transferred to other glyphs as well.

[irenevlachou]

not really any comments. Here is what I have done in my version, which is actually minor changes and some alternate designs to Emilios just to have a greater choice. Whatever the terminal design we decide, I'm in, I just thought to try a more spiky approach.

UC:

• “A” italic is 5units sorter than the upright and this is reflected on the design of Delta and Lambda
• I made an Omega italic based on my Omega upright

lc:

• alpha: new version follows a more normalised contrast
• beta: a more calligraphic approach in the middle stroke. Also I have straightened the rounding on descenders. I think we can treat tops and bottom, straight and round, slow and fast terminals differently.
• gamma: more round and narrow and straightened vertical stroke
• delta: fixed ascender height
• epsilon: a bit wider
• zeta: fixed ascender height and tried a new top that could go better with xi
• eta: straight terminal
• theta: fixed ascender height
• iota: a tiny too light
• kappa: a design closer to upright
• lambda: fixed ascender height
• mu: straight terminal
• xi: new top
• nu: like gamma
• pi: wrong x-height
• rho: straight terminal
• tau: wrong x-height
• upsilon: more round
• phi: a disconnected model, a much more boring choice than Emilios :)
• chi: more narrow
• psi: like upsilon
• omega: more open, shorter right side

[frankrolf]

Thank you both very much for your work, the PDF, and your comments. Here is my feedback – I tried to break it down letter by letter, and hope to be clear which variant I mean in any given case. Please don’t take any of it personally, and let me know if anything is unclear.

general:

• I like the straight descenders for βημρψ better – Source Serif doesn’t have much roundness in the details anyway, so I’d prefer keeping that through the Italics.
• I am noticing “fixed ascender height” as a comment in a few letters. The ascender height in the Greek is deliberately a little lower, to correspond with the shorter descender. Also, Greek ascenders have a different nature than Latin ascenders – while Latins are usually just sticks poking out, there is much more mass in the Greek. For that reason, I’d prefer “un-fixing” those changes.

α

The normalized contrast (Irene’s variant) makes it look like it is carrying a backpack. I prefer Emilios’ overall construction. You can find the same distribution of weight in gep, so I welcome this solution.

β

See general feedback. Bottom not too thin, and descending stem angular.

γ

I prefer Emilios’ variant because of the stronger relationship to v, however I’d also like to see a more angular descender. See upright.

δ

See general ascender height comment.

ε

The fundamental difference between the two approaches seems to be the openness and overall width. I don’t have a preference.

ζξ

Having two different top elements in ζξ looks odd to me – @thynem already addressed this above. I agree with him – Irene’s look more harmonized, however the top of the ξ needs some evening out – the follow-through looks broken. Also, should those terminals extend beyond the ascender that much?

η

Please don’t round out the areas where counters meet the stem, which does not correspond to the overall design.

θ

Irene’s variant: Italic angle looks off especially in text – top/bottom counters not balanced.

ι

In the context of other letters, the lighter variant looks better to me.

κ

The variant with the arched top part looks a little busy – I like Emilios’ suggestion better. It distributes the whitespace a bit more evenly.

λ

Same feedback as on the rest of the ascenders. Left leg looks a little light in both variants.

μ

No comments about the overall shape, see general descender-comment.

ν

See general comment about contrast

ο

Looks very close to the Latin although proportions have been changed. In the Roman masters, I also changed the contrast a bit – is that a good idea or no?

π

There seems to be a bit of a disconnect about the Italic angle. In text, the slightly-more upright version looks better.

ρ

Both good. Emilios’ has a little more weight on the left stem which balances out the letter – Irene’s has the better descender.

σ

No feedback from my side. I’d expect the flag to be a little longer, but you are the specialists!

τ

No feedback from my side.

υ

The right-side terminal curling back toward the main stem creates a bulged countershape. I think keeping it open is the way to go. Also, don’t make thins too thin.

φ

While I find the connected, heart-shaped variant charming, I don’t think it works. It changes the perceived angle of the letter and creates two unevenly-sized counters. Irene’s version is good in text.

χ

Making the proportions slightly narrower is a good idea, which helps the spacing. I like Irene’s version, however with thins a little beefier.

ψ

Same as υ – I’m not a fan of the back-curling terminal. Terminal in Emilios’ variant needs to be slightly bigger.

ω

Unfortunately I think I prefer the rationalized version, although the asymmetric variant is so much more flavorful!

ςζ

I don’t know how low the left bowl of the letter should “kneel” on the baseline – the curvier one (with less knee) seems to go better with the Italic angle.

caps

• I am a fan of Irene’s Ω, but I would like to see a heavier horizontal bar. Maybe lift the whole connection a little upward to make that possible? I think I made the same comment on the Roman master.
• Irene’s Ξ seems to be quite wide, much more so than the E (which is normal I suppose), but even wider than the Η? In any case, Irene’s variant has better horizontal spacing.
• Same spacing feedback on Irene’s Σ here.

---

I think we can kick of one more iteration of the regular masters (both Roman and Italic), and then carry over the design to start the light and heavy Italic masters from the product of this iteration.

Thanks for the great work everyone!

[thynem]

Updated versions according to the previous discussions are now on Github.

For a more detailed preview, character sets and running text, you can download the PDF: SourceSerifRomanItalicGreekTestDoc181119.pdf

Small addition:

• The character ϗ has been designed for the Italic and a unicode value was given.

[frankrolf]

Thank you for this excellent work! I have very few comments:

ζξ

The almond-shaped details are a little too round here. Usually those details are slightly elongates, see (Roman) afυν. Maybe the whole terminal can be turned in the direction of the stroke, so it basically is a pronounced swelling rather than a literal ball terminal.

υ

Same feedback as for ζξ. The ball terminal has been shortened here, and feels too abrupt for me now.

ϗ

The loop feels a bit un-elegant. Does it really need a ball terminal at the end?

Thanks as well for working on the upright! I like how it looks!

I think it’s time to transfer this work to the other (ExtraLight and Heavy) masters now, please let me know if you need any assistance with that.

Thank you both!

[irenevlachou]

Hey Frank, sorry for the delayed answer. We are good to go after your comments, and wrap up the uprights and move fast to the other italic weights. I guess next week will be a bit slow but I'll upload updates.

[frankrolf]

Thanks Irene! Looking forward to those commits – Happy Holidays!

[irenevlachou]

Hi @frankrolf I just uploaded new files for the Upright greek and a fresh Extra Light Italic. The Black Italic is on its way too. Here are some comparison gifs for the uprights and a PDF for the Extra Light Italic. Looking forward to your comments!

GRKMaster0uprightVsitalic.pdf

[frankrolf]

Dear @irenevl, thank you for your work! Here are a few observations:

SourceSerif_2

• ξ and ζ need re-spacing of their RSB, since the heavy mass formerly present in that area is missing now.
• δ is a nicer solution, but the joining of dark-dark elements also looks very dark in text.
• α has better/different weight distribution now, but also a heavy blob on top of the shape
• β – I wonder if the weight shift went a little too far here – the bottom bowl seems to balloon up now.
• ξ – the shortened top makes it look darker. Perhaps the connection needs some work?

SourceSerif_1

• ω – does it need to get a little narrower to compensate for the additional whitespace?
• Ω – the change of weight goes in a good direction, but the bowl feels a bit droopy now
• δ – some heaviness issues on the left side of the bowl
• ςζξ – I find the terminal slightly too pointy now.
• Φ – Is shortening the stem really better? The correction doesn’t look good to me in text, in a word like ΑΣΦΑΛΕΙΑ for instance
• Σ is better wider, and more balanced. Thank you.
• Ξ – the middle bar is not visually centered

SourceSerif_0

• Ω is now fully monolinear. I think it needs some contrast in the complex area around the bottom
• α – why so narrow? I agree it can be a tad narrower, but now it looks out of place in a word like σασί, καταστροφής, απορροφάται etc.
• ω – left side seems a little rounder than the right. On purpose, perhaps?
• Δ is now very narrow. The change is welcome, but I’d relax it a little and go slightly wider

SourceSerif-Italic_0

• Ω – same as above, but even stronger so. Top round shape seems uneven.
• Δ is also very very narrow here
• Π is too wide
• α – the weight on the top terminal seems a bit much and quite distracting. Probably the crossing needs to be thinned out, too
• δ – terminal hanging low
• β – seems too wide in relationship to other characters
• ζξ – have different angles, perhaps the ξ is not “Italic enough”?
• κ – no contrast in leg
• ω – drawing needs work
• ξ – top right is perking up, looks weird in some words like έδειξε. Top right is dark.

General

• When changing spacing, please also consider the -tonos glyphs. That makes it easier when checking real text.
• Please note that the ExtraLight Italic masters has very similar proportions to the _1 (text) master – the glyph widths don’t change much between the two. The weight only starts changing significantly between _1 and _2 masters.

Thanks again for all your work, much appreciated! I am looking forward to future developments in the Greek Italic (especially in master _2 ;-) !

[irenevlachou]

Hi @frankrolf, thanks for your comments. All files are now updated and the last SourceSerif-Italic_2 is uploaded. Looking forward to your comments! :)

[frankrolf]

Hi @irenevl, I had another look, and here is what I think. Please feel free to defend any of your decision, at this point I can mainly comment based on my gut feeling (and previous design decisions):

SourceSerif_0

The new α is too narrow:

---

The new ξ seems a little imbalanced in terms of whitespace distribution of the two counters:

---

How narrow should the Δ be? I feel it’s been relaxed here, but I’m not sure it is not still too narrow. Additionally, is the Δ heavier than the A?

---

The new Ω is a success, and I believe the previous problem (monolinearity) has been resolved. A bit widely spaced, perhaps?

---

This is minimal, but is the crossbar of Θ now slightly low?

---

Ξ now significantly wider than Ε. Is that OK?

---

Ψ looks more balanced. Thank you for this.

SourceSerif_1

Rounding off the counter of α makes it a unique element in the typeface – no other counter connection is rounded. Please backtrack a little bit.

---

New υ looks a bit narrow in this context:

---

New δ: nice!

---

σ now has the kink on top I tried to avoid. OK?

---

φ got wider, ψ narrower. While I have nothing against the former, the narrow ψ maybe looks a little out of place here:

---

Ω is quite a bit darker than surrounding characters:

---

Δ the width here is fine

---

It seems like the O-shape of Θ got heavier. Should this be the case?

---

Ξ I am curious about width (as above), vertical alignment, and weight of the crossbar.

---

Φ The removed overshoot bugs me in words like this one. Is this really so uncommon to have the bar overshoot in Greek? (The Cyrillic Ф overshoots in many designs)

SourceSerif_2

I noticed a normalization of the contrast in a more static (vertical) direction. What is the reasoning here? More darkness on the page, or general evenness?

The new α shifts the counter downwards but creates a big blob on top. Please consider enlarging the triangle of whitespace that cuts into the stem.

---

I like the new ω!

---

I feel something’s missing now from the bottom of the β. The connection to the stem is so much heavier – maybe the bottom triangle of whitespace should be enlarged?

---

The old δ wasn’t good, but the new one also has problems. Mainly the dark spot where the diagonal connects to the bowl. Also, I feel the whole letter is leaning to the right. When comparing δςδςδς I think there is too much of a discrepancy in heaviness.

---

ζ and ξ – see heaviness remark made right above

---

ρ – like β – I like to see more whitespace in triangle at the baseline to make the letter less static-looking.

---

Ω – I am surprised to see that the horizontal heaviness has completely vanished. Is this personal preference or just normal for heavy Greeks?

---

Δ – same question as Ω, although here it actually looks better! ;-)

---

Θ looks better wider, but this is a little extreme. Perhaps shave off something from the inside too and make the whole shape slightly narrower again?

---

Ξ – same questions as asked above. To my eye, it mostly just looks very wide now. (However I can see that many macOS system serif fonts also have a wide Ξ)

---

Φ still uses the overshoot as discussed above. If it needs to be removed, let’s do it consistently (but I think it’s more even with overshoot in a situation like this):

---

Thank you for the better Ψ.

[frankrolf]

Some brief notes about the Italics:

SourceSerif-Italic_0

Generally narrow. The Latin Italic_0 is not narrower than the Italic_1 master. β left side heavy θa bit angular? κ this construction makes it look “less Italic” than the Roman ξ detailing too complex π terminal too heavy. compare to c. ρ narrow? σ also angular (bottom left) υ dark on left side χ no contrast ψ quite flamboyant, which creates uneven counters. Heavy left side. ϗ needs more love. Is the terminal needed?

SourceSerif-Italic_1

α different angle compared to other letters δ terminal very curly ζ and ξ – detailing too complex υ right side not as round as in master _0, which makes the letter look out of place χ very aggressive ϗ needs more love.

[irenevlachou]

Hi @frankrolf I updated the uprights and here are some gifs with before and after and my comments:

SourceSerif_0: α: I overdid it, I was trying to avoid the balloon shape that it had before :)

ξ: I lowered a bit the middle bar and made the horizontals a bit longer. I also smoothed the outstroke at the bottom and applied the change to ζ and ς

Δ: it should look as wide as Alpha and Lambda, so it needs some optical compensation. Ιt was indeed 2 units heavier than Alpha, well spotted!

Ω: fixed the spacing

Θ: you think? It really bugged me before I found it a tiny bit too high and not optically centred :)

Ξ: it is supposed to be much wider than Epsilon but it was maybe a bit too much. It looks different than before because the length relation between the top and bottom horizontal strokes compared to middle one has also changed. I made it a few units narrower though to match the previous design.

SourceSerif_1:

α: oops, fixed

υ: it wasn't really much narrower just narrower opening. It looked too wide and open before and it was distracting in text all this white space. I made the left side a bit more upright based on the your drawing and the right a bit rounder. More balanced now? The same change applied in ψ.

σ now has the kink on top I tried to avoid. OK?: done

ψ a bit wider

Ω less heavy

Θ: the O shape was lighter before and there is no need for optical correction in the Regular weight I think

Ξ is updated

About Φ, I have kept the top overshoot across all weight and styles. I just removed the bottom overshoot so we can keep the baseline alignment. Are you referring to that?

SourceSerif_2:

The whole idea wasn't to normalise the contrast, but make it a bit more even across the weights. The ductus was changing direction a lot between master 0 and 2 and the designs looked unconnected between them to a certain extent. The new proposal is to even out these differences without ruining the character of the greek that has been chosen for this design. To even out the big differences in a sense.

β and ρ fixed

δ fixed

Ω is usually thinner but this one was too much. Now Δ, Ξ,Σ and Ω have balanced heavy horizontals.

ΘThat was way too narrow before, in the heavier weights we tend to make it wider than O, but probably was too much

Italics will be ready by tomorrow.

[frankrolf]

Great work! Thank you so much for considering my comments, and explaining why you made certain modifications.

The shift of contrast you mention is well-spotted, and was in fact intentional. You can try and guess who suggested this idea, but I don’t really think it makes logical sense and/or works out, so normalizing the contrast across masters is a good idea. I am looking forward to the Italics!

[irenevlachou]

Hi @frankrolf the two italics are updated according to your comments. Let me know if you have any comments for the third master.

SourceSerif-Italic_0:

β left side heavy

done

θa bit angular?

yep it was a bit

κ this construction makes it look “less Italic” than the Roman

I think it's better to avoid something more complex than this. It's calligraphic enough and for an italic it does represent a fast writing. Do you have something specific in mind?

ξ detailing too complex

And the opposite here :) usually zeta and xi are just slanted versions of the upright. So we thought that would be more interesting if it had this curly top to give the italic a distinctive note. It's also convenient because zeta and xi are rare letters.

http://letterfrequency.org/letter-frequency-by-language/

π terminal too heavy. compare to c

I thined that down along with other similar terminals

ρ narrow?

done

σ also angular (bottom left)

ok

υ dark on left side

ok

χ no contrast

ok

ψ quite flamboyant, which creates uneven counters. Heavy left side.

aaa the famous greek temperament :D I tried to make it a bit less theatrical

ϗ needs more love. Is the terminal needed?

Not necessarily, you had it in y so I thought to give it a try :)

SourceSerif-Italic_1:

α different angle compared to other letters

done

δ terminal very curly

I made the terminal a bit shorter and more upright, is that what you meant?

ζ and ξ – detailing too complex

same as in uprights

υ right side not as round as in master _0, which makes the letter look out of place

done

χ very aggressive

it makes an aggressive cat sound! Any smoother now?

ϗ needs more love

it had its share!

[irenevlachou]

Hi @frankrolf masters 1 and 2 have been updated and have complete sets now (archaic symbols).

[frankrolf]

Great, ευχαριστώ! I expect I’ll find a chance to give feedback this week.

[frankrolf]

Source Serif 0

α: width is better. Please make sure the new advance width follows along with the accented glyphs.

---

β is narrower in all styles, which I think is a good change. This particular β got a bit light at the 1–2 o’clock section of the bottom bowl.

---

ξ – is the middle stroke too long now?

Source Serif 1

αδυς – those changes look good!

---

σ – good kink-avoiding work!

---

ψ – is the vertical bar slightly off-center now?

---

Now the Ξ is even narrower than in the original version. I don’t know what to believe anymore! 😂

Source Serif 2

The ρ suddenly has additional spacing on the right side. Not on purpose, I assume? The α has been fixed! Thank you!

---

I am curious about the tonos accent. Where should it be?

---

New δ – nice!

---

New Ω – also great. I think it works!

---

I think the new Δ is also good. In the process, the left hairline got a little flare, which I think needs to be corrected:

---

It’s nice to see this progression. Works!

---

Ξ got slightly narrower but also thinner. Is that OK?

Italics

α – the weight progression seems to be a little odd. What is thick and what is thin? Master_1 seems monolinear almost (except for the final upstroke) Master_2 also has some weight issues in the top-right terminal.

β – different construction logic across masters

---

θ – the shape is very different across masters

---

κ – Master 1 has calligraphic flare, the others don’t

---

λ

The weight distribution for the extra heavy Italic λ creates a bit of a strange shape, not seen in any of the other λs. Is a solution like in the Roman Heavy master thinkable?

---

χ has a bit of a drawing problem – especially in Italic_2. Why do the heavy stroke terminals need to be angled in this almost-horizontal way? In the Roman, I tried to make the overall problem less severe by going easier on the lead-in hooks (top-left, bottom-right).

---

Ξ – decisions taken in Roman need to be carried over to Italic

---

Φ still has phantom components in Master_1

---

Ω seems heavier on the left than on the right (in 0 and 1 masters):

---

Tonos angle consistency

Source Serif Italic 0

The biggest change seems to be the wider ρ, which is a step in the right direction. I still have a feeling that the overall design is too narrow. Look at the line lengths of those screenshots to compare masters _0 and _1:

In short: I think the whole design needs to be wider.

Source Serif Italic 1

koppa stigma digamma sampi are missing from this file.

α: quite heavy TR terminal

---

ζξ I made a comment about the general business of those shapes, and I am still feeling they are too complex. Instead of the ball terminal, could we try a solution similar to the α-terminal? Basically just a stroke that tickens?

Source Serif Italic 2

Thank you very much for your work on this master! I think it goes in a promising direction, but there is not a lot I can comment on right now. Some things are covered in the general review above – one general thing to say is that I try to avoid heavy blobs as much as possible. Right now, I see βεξλϗ in blob-territory. Another thing I’m noticing is the consistency of thin terminals. For instance, υ and ω – I like the mass of the ω-terminal better. Consider not bending down those terminals too much, then they are easier to balance (as mentioned above in my note on the χ)

Minor: the counter of theδ needs work:

---

Thank you ever so much for your great work! I am glad I could get these comments back to you this week, because I’ll be teaching in the following weeks, and I won’t have so much time for feedback. Nevertheless, feel free to ask questions, and – of course – to update the files! ;-)

[irenevlachou]

Hi @frankrolf I just uploaded new files for the Upright and Italic greek with all small adjustments. The main updates are the the overall reduced width of the Italic_0 and the simplified top of the Italic ζ and ξ. Looking forward to your comments!

[miguelsousa]

Just chiming in to comment on one detail that jumped out to me:

• In the Black italic Theta, the serifs of the crossbar look thicker than in the roman.

[irenevlachou]

Thank you @miguelsousa , well spotted. All fixed and updated

[frankrolf]

Dear Irene, thanks again for all your work! I was a bit scared when you said the “overall reduced width of the Italic_0”, but I am glad you increased it as suggested!

• Please note that koppa stigma digamma sampi are still missing from Italic_1.

• In almost every style, the capital-letter + tonos combinations have been changed, sometimes significantly. What was the reasoning behind this? I created the tonos in a way so it would sit on the left margin of the glyph, which IMO is a good compromise between visual distribution & too many kerning pairs. See these screenshots (red is old, blue is new):

I think that the new spacing of the tonos for Omicron and Alpha in styles _0 and _1 is a bit tight for a text face. The Alphatonos combination in style _2 is likely a mistake. On top of that, the tonos.cap component is no longer compatible across styles.

I can fix all that, but please let me know what your thinking was – I assume it is just an execution of the “tonos angle consistency” complaint from above.

If you don’t mind, I would like to do an editing pass by myself for now, and send the result back to you. Sounds good?

[irenevlachou]

Hi @frankrolf, the Alphatonos combination in style _2 is a mistake, but the styles _0 and _1 aren't. I found that the distance between the tonos and Alpha in Alphatonos and the rest of the accented UC is too different, and I do prefer to have this distance optically the same. About the spacing of the accented UC, it's something to discuss I guess and it relies on personal preference. I see that you try to space them somewhere in between, but in some cases that doesn't really work I think (the space in your example in the black weight is almost eliminated between the last letter of the first word and the first accented letter of the second. Also how does that work with quotes and brackets? The classical way is to space these things exactly in the width of the UC and have the accents hanging. It creates loads of kerning pairs etc but it's more efficient for classical typesetting greek without creating rivers in the left side of paragraphs.

[frankrolf]

@irenevl I’d be curious to try out a few different approaches. Do you have a sample text in which the rivers you are describing would be observed?

[irenevlachou]

This sample text has sentences starting with accented polytonic caps. One example is set with Minion Pro and the other in Minion 3. In Minion Pro the accents hang outside the box, which in my opinion, gives a more solid text box, and in Minion3 the text doesn't really look aligned on the left. I'm attaching 2 pdf's that you can grab the text and two zoomed screenshots. Imagine poetry typesetting when the lines are fewer and misalignments are more obvious. Minion3.pdf MinionPro.pdf

[frankrolf]

That’s a good visualization, thank you @irenevl ! Thanks also for the text. I’d like to point out that InDesign’s “Story” panel will help alleviating the problem overall, but of course it doesn’t create a visual left margin that uniform. I have checked Source Sans Pro, and it uses the same approach of half-spacing the hanging tonos, consistency across the families is another fact to consider. However, I will bring this up in our Design Meeting, and ask Robert why he chose to go with half-hanging accents in Minion 3.

[frankrolf]

Hi @irenevl, I have completed my round over the Greek lowercase.

I made only very few fundamental design changes (κ), but many little adjustments – I mostly straightened up the design a bit wherever I found it too curly. Also, I changed the curve structure in some cases. Rather than describing all the changes I made, here are three PDFs:

Source Serif Pro Master - Greek before.pdf Source Serif Pro Master - Greek after.pdf Source Serif Pro Master - Greek difference.pdf

Please let me know what you think of these edits, I am sure there are things that still can be adjusted.

[irenevlachou]

Hi @frankrolf , thank you for your PDF's and your work. Generally I'm fine with the changes, I prefer k's without in-strokes (or serifs?) but since we didn't change it on the upright it make more sense to keep it here.

just a few notes: α I'm not sure I agree with the new in-stroke; it's nice and balanced but we don't see this kind of terminal anywhere else. We had it before the same as in κ since both had the same slant. Any reason to change it? The counter on master_2 looks really squashed.

β in master_1 the top counter becomes too thin just before the junction, if you compare it with the upright.

The horizontal stroke of θ in masters 0+1 isn't the same weight as in the upright masters.

The λ in master_2 it doesn't have any diagonal stem left as it had before and somehow I feel it needs it. In the upright there is a tiny straight part that give the letter some kind of stem that can "hold" the curly in-stroke and the branching of the left diagonal.

In all three masters of ξ, the middle bar has been moved lower, making the top counter bigger. Any reason for that? Also now they aren't in balance with the upright styles.

π in master_0 isn't the left terminal too thin now compare to the upright?

Accent positioning: thank you for updating the uprights, for somer reason I never manage to upload successfully my updates.

some comments: The position of tonos in ά I think is too much to the right. It shouldn't be optically centred to the whole letter as just "pointing" to the highest point of the bowl.

Same with the ή, the tonos should "point" at the junction. In master_0 should be a bit to the left.

Let me know what you think and I can quickly update the files and add the missing symbols from master_1.

[frankrolf]

Hi @irenevl, thanks for the feedback!

α I'm not sure I agree with the new in-stroke; it's nice and balanced but we don't see this kind of terminal anywhere else. We had it before the same as in κ since both had the same slant. Any reason to change it? The counter on master_2 looks really squashed.

I admit this is α is an experiment. I used that kind of terminal thinking of πστ, but it may just be the wrong way of looking at it. Using the solution of the k made the whole letter too soft. The counter was changed because the shape was lacking contrast:

β in master_1 the top counter becomes too thin just before the junction, if you compare it with the upright.

That’s true.

The horizontal stroke of θ in masters 0+1 isn't the same weight as in the upright masters.

I think I just aligned it with the overall weight of the design – since the Italics are slightly lighter than the Romans, I didn’t think much of equalizing them. However, I think I went overboard with master _1.

The λ in master_2 it doesn't have any diagonal stem left as it had before and somehow I feel it needs it. In the upright there is a tiny straight part that give the letter some kind of stem that can "hold" the curly in-stroke and the branching of the left diagonal.

Fair. Do you think the flag can curl down more to facilitate that kind of solution?

In all three masters of ξ, the middle bar has been moved lower, making the top counter bigger. Any reason for that? Also now they aren't in balance with the upright styles.

I did that to balance the top and bottom counters better. In the Italic, the bottom heavy stroke (which is almost horizontal in the upright styles) bends down quite a bit, creating a big white space. I think I can backtrack in masters _0 and _1, but possibly not _2?

π in master_0 isn't the left terminal too thin now compare to the upright?

This is the terminal of c. The previous one very visibly popped out in text, and this helps it a bit. If you want to see it a tad heavier, I’d make it 2-3 units but not more.

Accent positioning: thank you for updating the uprights, for somer reason I never manage to upload successfully my updates.

The uprights were updated only slightly – the Italics more than that. I got some feedback regarding the spacing of hanging accents from @tiroj about the placement of those:

The default spacing of the uppercase diacritic characters has to accommodate their use in the middle of a sentence and preceded by another word and a space — e.g. in names, titles, etc. —, and not just at the beginning of a sentence or the beginning of a line. That can be done either in the sidebearings, as I've done in Brill, or by positively kerning the diacritics woth hanging marks off the word-space glyph. I avoid the latter mechanism because kerning, especially involving the space glyph, is unreliable in some software.

I think this makes sense, and since we have to expect situations without kerning (MS Word), I’d like to go for this solution. John has more to say:

The handling of the accents and breathing marks at the beginning of a line is sort of an independent question from the general spacing of the diacritics. There are lots of examples of Greek typesetting that use fully hanging diacritic marks, as Irene illustrated, often even pushing the marks on Alpha out into the margin. This suggests using something like the optical bounds GPOS features to specify the distance to hang diacritics, or even decomposing some of the cap diacritic combinations contextually in order to handle the diacritic marks independently of the letters.

I can investigate how to do that.

The position of tonos in ά I think is too much to the right. It shouldn't be optically centred to the whole letter as just "pointing" to the highest point of the bowl. Same with the ή, the tonos should "point" at the junction. In master_0 should be a bit to the left.

Good to know! I guess we need to wait for the optimal shape of α before the tonos can be placed correctly. FYI: anchors above the letters are already in place, so if you move the tonos, please move the anchors by the same amount. This may help: https://gist.github.com/frankrolf/ec5a50aaad6d7b3b471498e39ebe3b4d

Please feel free to make any edits, I am looking forward to seeing them!

[irenevlachou]

Hi @frankrolf thank you for your reply, I've updated the files and uploaded them. I tried to make the changes very discreet and please both of our tastes :)

@tiroj comments were very useful and I totally agree. I just think that it is as important for these characters to sit as nicely in the beginning as in the middle of the sentence. The optical bound solution would be very convenient since it is a much cleaner solution and it'll save us from all that kerning. We are taking about the opbd feature, correct? If only the opbd feature was supported, things would be great! Until that happens, my spacing approach (and that of Minion Pro, which many Greek typographers know) seems to be the only simple way to get left-alignment at the start of lines. So the reality seems to be a question of weighing up the benefits vs the non-typographic app's support.

[frankrolf]

Thank you @irenevl !

I tried to make the changes very discreet and please both of our tastes :)

😂 It’s almost difficult to see new changes now – which may be a good thing! Do you think we are about to wrap this up? I assume the α was OK after all then? Please let me know how you’d normally proceed at this point.

One note about the koppa stigma digamma sampi etc in master 1: Reading the commit history it seems the .glif files existed already, but were just not in the contents.plist file of the glyphs directory. How did that even happen? Sorry if this was an oversight on my end.

[irenevlachou]

Hey @frankrolf

all good with α Ι think it's looks nice at the end, it has a fresher look :) I just adjusted a bit the counter, but nothing major. Kerning is next. I'll upload any updates as soon as I have something ready.

[frankrolf]

Thanks @irenevl ! Parade make sure you use the latest UFOs – the greek-italic branch no longer exists, and has been renamed to dev. I have created Greek kerning groups already. Since we’ll be working on the same files, it would be good for you to work on your own branch – I’ll make greek-kerning for you.

[irenevlachou]

Hi @frankrolf kerning is done in all three masters, please have a look and let me know if any changes needed.

[frankrolf]

Will do. Many thanks!

[frankrolf]

Thank you so much for your help @irenevl and @thynem – Version 3.0 was released today! https://github.com/adobe-fonts/source-serif-pro/releases/tag/3.000R