Hinting and font rendering

Hello, i 'll give yall a short explanation.

I'm designing a typeface with an intent to publish it and be usable in a voriety of plataforms but...

Figma and Microsoft Edge (Chromium project) render them differently with chrome/chomium being extremely bad. 

Just for demonstration:

On Figma at 15px



On chrome/Edge chromium at 15px









«1

Comments

  • Mike DugganMike Duggan Posts: 237
    how is the hinting done in the font? autohints? you might want to check the GASP table, you can find much more information on the gasp table and hinting here in this document I wrote https://googlefonts.github.io/how-to-hint-variable-fonts/
  • John HudsonJohn Hudson Posts: 2,973
    In addition to Mike’s advice, I would also recommend respacing the typeface: it is much too tightly spaced for use at the size you demonstrate, and this is contributing to some of the rendering problems you are seeing, as well as to crowding that diminishes legibility. The space between the letters and the space within the letters need to be more balanced, so that each letter is distinctively recognisable and the antialiased rendering is not blurring them into each other.
  • DiegoSouzaDiegoSouza Posts: 38
    In addition to Mike’s advice, I would also recommend respacing the typeface: it is much too tightly spaced for use at the size you demonstrate, and this is contributing to some of the rendering problems you are seeing, as well as to crowding that diminishes legibility. The space between the letters and the space within the letters need to be more balanced, so that each letter is distinctively recognisable and the antialiased rendering is not blurring them into each other.
    Thanks for the answer, i'm Just starting to adjust those things but still some drawing issues, i'm seriously thinking about restart some glyphs. 

    When spaced spaced It resembles univers 45 light, but my target was towards something like neue Haas, nimbus sans ou PP neue Montreal.
  • DiegoSouzaDiegoSouza Posts: 38
    how is the hinting done in the font? autohints? you might want to check the GASP table, you can find much more information on the gasp table and hinting here in this document I wrote https://googlefonts.github.io/how-to-hint-variable-fonts/
    Yes, It is autohinted, i'm searching How to manually hint but from the things that i found i didn't know How to improve It since bluevalues, stemhw and stemvw seem correct i tried to play with blufuzzmfrom 1 to 0 but didn't like the result
  • Mike DugganMike Duggan Posts: 237
    perhaps ttfautohint and generate as a ttf? or VTT for manual hinting

  • Mike DugganMike Duggan Posts: 237
    the first image is very odd, the spacing as John noted is very very tight, the word 'will' for example, it looks like it is rendered using older gdi compatible widths. If you want to check the gasp table and report back, this is a ttf right? and or send me the font I can take a look at it, and see If I can replicate any of this rendering 
  • John HudsonJohn Hudson Posts: 2,973
    Yes, It is autohinted, i'm searching How to manually hint but from the things that i found i didn't know How to improve It since bluevalues, stemhw and stemvw seem correct i tried to play with blufuzzmfrom 1 to 0 but didn't like the result.
    Oh! So you are making CFF OTF fonts, not TrueType?

    Regarding spacing:
    my target was towards something like neue Haas, nimbus sans ou PP neue Montreal
    Those are Helvetica-relatives or derivatives, and none of them were designed or spaced for text sizes or for reading on screen. They work great in larger sizes where their tight spacing produce strong graphic shapes, but for text readability sans serif types need to use a different, more generous spacing model.

  • DiegoSouzaDiegoSouza Posts: 38
    Yes, It is autohinted, i'm searching How to manually hint but from the things that i found i didn't know How to improve It since bluevalues, stemhw and stemvw seem correct i tried to play with blufuzzmfrom 1 to 0 but didn't like the result.
    Oh! So you are making CFF OTF fonts, not TrueType?

    Regarding spacing:
    my target was towards something like neue Haas, nimbus sans ou PP neue Montreal
    Those are Helvetica-relatives or derivatives, and none of them were designed or spaced for text sizes or for reading on screen. They work great in larger sizes where their tight spacing produce strong graphic shapes, but for text readability sans serif types need to use a different, more generous spacing model.

    the first image is very odd, the spacing as John noted is very very tight, the word 'will' for example, it looks like it is rendered using older gdi compatible widths. If you want to check the gasp table and report back, this is a ttf right? and or send me the font I can take a look at it, and see If I can replicate any of this rendering 
    Again, thanks for the answer and Sorry for the bad typing, i'm using my tablet without english installed in It, so sometimes It is correcting to portuguese.

    Yes, it's a ttf file with autohint, i was using cubic splines but converter to quadratic ones. Tonight (at least for me) i'm going to try fix The spacing, If nota being able i Will responde here and contact you Mike 


  • DiegoSouzaDiegoSouza Posts: 38
    Updates, i still adjusting the gasp tables/understanding how it works but i can some positive results.



    I also started to to some manual hinting in some spot that fontforge started to fail specially u. 

    Kerning still being performed

    If you yould like to give some critics towards the shapes itself, thanks

    as i said before with more spacing it resembles univers 45 light and what i found more interesting was that i never intended this visual similarity 




  • Mike DugganMike Duggan Posts: 237
    it has changed in appearance, but is now unhinted
  • DiegoSouzaDiegoSouza Posts: 38
    it has changed in appearance, but is now unhinted
    Again, thanks for the answer and the knowledge given.

    Well, i saw this message and i got confused, it's technically hinted, autohinted, it was conveted to quadratic splines but i had to convert back to cubic because i was facing a lot of trubles trying to draw shapes. I still fixing some things but trying to perfom a "final" review on the lowercase letters. 

    still experimenting with gasp table but now i start to think about this message again, in certain ppi the quality of the drawing falls dramaticaly, specially at 18px p/em... i go back to font forge, clear hints, play again with gaps table and autohint again and basically nothing changes.

    i'm thinking about manually hint everything but form what i found online it isn't necessary because as i said bluevalues etc... seems correct.

    in font forge i clicked on "optime for cleartype" can you give me any light ? 
  • I’m trying to say this as nicely as I can: you might have hinting issues, but your main issue that your shapes are extremely poorly drawn and a far cry away from any final shapes that would lead to a well-functioning design.

    I’d recommend taking a week away from the typeface and revisiting, it will be easier for you to see mistakes. Start with comparing and unifying things like stroke weights, widths of comparable letters (see for example c/e/o), curve tension, styling of stroke terminals (like r/f/j/t/y).

    Long story short: don’t worry about hinting. Worry about the drawing.
  • John HudsonJohn Hudson Posts: 2,973
    edited June 2023
    as i said bluevalues etc... seems correct.
    Things like bluescale and bluezones are specific to Postscript/CFF hinting, and are not directly used in TrueType hinting. This may get confusing, however, because some TTF autohinting may make use of bluezones in some font tools as analogues to TTF control value alignment zones. It is also possible that when you say you autohint the font you may be applying PS/CFF autohinting that isn’t applied at all when you output a TTF, which would explain why ‘basically nothing changes’ when you re-autohint. Without know what what tools and methods you are using to output and autohint the font, we’re sort of guessing here.

    The gasp table is specific to TrueType, and determines how the TT rasteriser interprets and applies instructions with regard to gridfitting and antialiasing. You shouldn’t need to re-autohint after adjusting the gasp table, because the whole point of gasp settings is that they are set independent of the instructions themselves and only affect how they are interpretated (and sometimes ignored) by the rasteriser.

    it was converted to quadratic splines but i had to convert back to cubic 
    What you do with outline format in your tools is independent of what gets written to the output font. TTF uses quadratic outlines, and CFF uses cubic outlines. Tools typically convert to whichever format is appropriate when the font is exported. Most people work in cubic outlines in the design, and typically would work with quadratic outlines if doing manual TT hinting.

  • DiegoSouzaDiegoSouza Posts: 38
    I’m trying to say this as nicely as I can: you might have hinting issues, but your main issue that your shapes are extremely poorly drawn and a far cry away from any final shapes that would lead to a well-functioning design.

    I’d recommend taking a week away from the typeface and revisiting, it will be easier for you to see mistakes. Start with comparing and unifying things like stroke weights, widths of comparable letters (see for example c/e/o), curve tension, styling of stroke terminals (like r/f/j/t/y).

    Long story short: don’t worry about hinting. Worry about the drawing.
    Well that's my second try trying to design something and yes i agree with you, there are things to be finished, but that last image above doesn't represent the actual stage anymore, i already fixed the w v and y, a width problem with the o and m and recently found an option that i can control curve tention way better. Again, i still learning and i appreciate any tips. 

    But if i could ask you something would be for better directions, you're talking on a really high level of type design proficiency and that's certainly not my case, and many more people that come around here, so what i'm trying to say is be more clear... "the c is wrong, the curve of the b is too much oblique" or just give a reference to read, articles, an example of typeface to look and compare, errors that you committed in the past and many more. 

    For me as an example, i found extremely hard to drawn the bottom part of the c, U,  e, both bowl of the s and S and barely touched the ? 

  • DiegoSouzaDiegoSouza Posts: 38
    My output file is a ttf file, i'm using fontforge and using it's autohint function as taught here:

    FontForge -- An outline font editor for PostScript®, TrueType and OpenType fonts

    But some considerations before:

    I found an discussion online in the subject and they said basically what i've said, fontforge atomatically converts everyting but they keept insisting that it should be converted to quadratic splines and again i found it to be a really bad idea.

    Also, when using cubic splines on fontforge you can autohint but can't  instruct to instruct you have to convert to quadratic... 

    So what I understood was that all this time was autohinting again and again for no reason ? OMG 

    well, learning from the pain 


  • You could look at online resources like OhNO Type School (paid but see their Instagram for I believe most of the same content for free but not in one single place), the GT Academy (I am part of GT), books like Designing Type (2nd edition), etc. but the main path to become a proficient type designer is simply to design a lot of type.

    I’m telling you about these "high-level concepts" because internal consistency across many different aspects is one of the biggest things to watch out for, and that is relatively easy to learn because it’s about checking and re-checking. Your design isn’t at a point where telling you about specific little bits being wrong would be appropriate.

    Type design is about learning to see what’s wrong, and learning how to fix that. Have fun learning! Just don’t expect your first few attempts to amount to much.
  • Dave CrosslandDave Crossland Posts: 1,391
    Is there more to read about GT academy than those posts? I have some folks who may be interested in joining if it's open to new members
  • Is there more to read about GT academy than those posts? I have some folks who may be interested in joining if it's open to new members
    I’m not sure what you mean. It’s a series of IG posts that will eventually be expanded to other formats (an actual website, maybe a book) — but it’s not classes or something like that.
  • Dave CrosslandDave Crossland Posts: 1,391
    Oh I see! haha I was confused by the comment on the first post, "Every Tuesday we share a new lesson with tips on creating a typeface from the ground up." I thought this implied there was more to the lesson than the post XD Sorry for misunderstanding :)
  • DiegoSouzaDiegoSouza Posts: 38
    Small update, thanks for the insights, well i'm giving a time to my typeface project and regularly will be posting more updates asking for more tips and insights, thanks for the time and patience 
  • DiegoSouzaDiegoSouza Posts: 38
    edited July 2023
    @Thierry Blancpain Already an update, i bought Ohno Type School and it's amazing, super recommended. I also checked GT academy and it's really interesting but i'm not the type of person that likes to use social media but i'm going to watch anyway. But i'm surely interested on buying the printed book, if your wish any time come true please be sure to send me a message as a reminder 
  • DiegoSouzaDiegoSouza Posts: 38
    UPDATE !!!! 

    after giving it some time... that's the result

    the yellow underline is for letters that i didn't ajusted yet

    i can see some improvements but i woud like some opinions of the professional eyes








  • DiegoSouzaDiegoSouza Posts: 38
    outlines


  • check out the stroke by gerrit noordzij, and the modular brushstrokes in the origin of the serif by edward catich if you havent already. I think if you have a better handle on the pen and brush strokes that many typefaces emulate, you might have an easier time spotting issues like the bowl of your /a becoming pinched, or the unusual swell on the curve of the /f, etc. this will save you time on your next font too. if you get a lot of feedback until it looks right and try another genre, what you have learned will not necessarily transfer over to that style. whereas if you start from a foundation in lettering you will be able to. a low contrast sans serif like your typeface is actually pretty difficult to learn with IMO, because the pen-like gestures are so subtle (a classmate once described a sans as having "nowhere to hide" which I think is accurate)
  • and we have a new update, after giving it some time i started to percieve some problems, some big problems some bigger than 2008 crisis... well i made adjustments and there are things that make me look to the other way about it. I have some new concerns, how to spot what's a drawing problem and what's a visual corrections looking some typefaces like "Neue Montreal, Sequel Sans ad Futura" i started to spot several curves and adjustments that look bad and some that are awful to me, but i'm trying to put in my mind that the problem is me and my lack of knowledge (and probavly it is).

    Also another update i bought karen cheng's book but amazon didn't delivered yet 

    Let me know what yall think about it!

  • hmnu get slightly “pinched” at the top right of h m and n arch (bottom left on u). Either need to thicken on the outside or the inside or a tiny bit of both.

    In your rounds on bcdgopq it looks like the vertical part needs to be heavier. Probably it is mathematically the same thickness as the horizontal part, but it needs to measure a tiny bit heavier to look the same. This and some other interesting issues are covered in https://www.youtube.com/watch?v=LR-CG5eB3nQ

    The same problem on the lowercase t, the crossbar LOOKS heavier than the vertical. Quite possibly they are the same, but if so, that is not what you need.

    The x is done with one diagonal heavier than the other, but perhaps a bit more than you need for visual balance in an otherwise monoline typeface. Whatever you do with the x, that treatment needs to be reflected in kvwy, and currently it is not, I don’t think.

    (w is a bit more complicated, there is more than one plausible scheme for distributing weight across the four stems. However, it does not yet follow any plausible scheme.)
  • In addition to what Thomas Phinney said: The lower left of the /a also seems a bit too thin to me. The /s seems to lean slightly backwards (and the /c and the /e to a lesser amount too). The dots of /i and /j are a bit too heavy. The tail of the /y joins in a strange angle to the upper right diagonal.
  • @Thomas Phinney @Linus Romer thanks, thanks and again thanks for the tips and material. 

    Well linus the s has been a problematic letter to me and indeed it's leaning backwards, i'm trying to fix that but it's a work in progress, about the tail well it was intentional i think it looks cool. 
  • Sander PedersenSander Pedersen Posts: 33
    edited August 2023
    Sometimes it helps to rotate (and maybe also slant) the whole /s one degree at a time between two upright letters until the letter looks more balanced, or it becomes more clear which parts needs to shift where. You can then redraw the letter over that as a base.
  • DiegoSouzaDiegoSouza Posts: 38
    edited August 2023
    update: 

    since kareng cheng's book delivery has been delayed i started to work on the capitals. In the begging i tought they were not great but ok, now i see them being really rustic. 

    just for demonstration



    and yes the letter w are 2 v's merged, on the right side it's a reworked B 
Sign In or Register to comment.