Kerning for beginners

AbiRasheedAbiRasheed Posts: 92
edited July 2016 in Type Design Software
Any good resource material for kerning in FL or in general about kerning pairs but for beginners? I somewhat understand how classes work in FL to some degree but  not enough to take advantage of anything else FL can do wrt kerning as I haven't gotten there yet & I'm hoping to get some info on other methods. Also is kerning affected based on the text size or is that something to do with the display? What I mean is like say if you go into kerning mode & start kerning at a text size of 128, but somewhere in between if you decided to change text size to say 24 or something then it starts to look very different which I think is expected as it may look tighter at small sizes. But the weird thing is even for shapes that are similar and have the same kerning value, one may end up looking tighter than the other when you switch text sizes, however the larger the text size the more accurate they look as intended. link to [gif] as an example, at 36, the first NH vs the 2nd NH look different in spacing though they have the same value but as I size it up it starts to look more uniform & similar.



  • Chris LozosChris Lozos Posts: 864
    The dreaded tracking comes into play as the user gets farther away from your target point size.  I tend to space a bit tighter than the target point size with the assumption that tracking out will not get you into as much trouble as tracking in [negative tracking]. This varies with weight as well, the bolder the weight, the quicker you can reach the excess tracking zone.
  • I think what you're referring to is simply Fontlab's poor rendering at small sizes. When you check spacing/kerning in the metrics window, don't base your decisions on anything under 50 points (I usually work at around 100 pnts for starters). Then, once it looks okay in Fontlab, make extensive printouts at high resolution to double check.

    Also, from your GIF it seems you're using kerning where you should be using spacing instead. BEFORE you even think about kerning (the yellow fields), it's important that your spacing (the white fields above the yellow fields) is spot on. Good luck!
  • AbiRasheedAbiRasheed Posts: 92
    edited July 2016
    Thanks everyone. I think I sort of understand what you guys are saying, that it boils down to good spacing which would make kerning easier. 
    @Jasper de Waard ah okay, thanks for clearing that up. Also yeah I kept the sidebearings at 0 because what I was working on needed to be kept very close and felt I could just instead kern it but I see your point though, thanks! @Dyana Weissman I've been doing exactly that and when I crossed 700 pairs or something I realised I had to get this cleared up because I felt it was just a painstaking inefficient process and that there might things you guys know to do it more efficiently and less time consuming. Classes are kinda nice, I only played with it a little bit but I'm taking note of what you said, Cheers! @Terminal Design that's very interesting what Ed Benguiat said, I never thought of it as 3s, but that totally makes sense.
  • Dyana WeissmanDyana Weissman Posts: 108
    edited July 2016
    If it helps, a 256 character set with good kerning would have 3,000 - 5,000 kerning pairs (with some exceptions depending on the design of course). 
  • In case of of camel case, lowercase, uppercase, lowercase. 
  • If it helps, a 256 character set with good kerning would have 3,000 - 5,000 kerning pairs (with some exceptions depending on the design of course).
    I think that are way to many kerning pairs for such a font. If you have more that 1000 there is something wrong with your spacing.

    There where some comments about the importance of kerning. I like to add one more: Up until the 70ties, maybe even later, there was no kerning at all in reading sizes and only really in headlines. I don't saying that you should not kern at all but make sure that font works without it.
  • Paul van der LaanPaul van der Laan Posts: 156
    edited July 2016
    I think that are way to many kerning pairs for such a font. If you have more that 1000 there is something wrong with your spacing.
    I think Diana meant a 256 character set from the good old days of Type 1 PostScript fonts, where class kerning was not possible. In those fonts you had to kern each glyph individually (in editors like Fontographer, or Robofog, or FontStudio). It was very common to end up with a kerning table with more than 3,000 pairs if you took care of all accented glyphs too.

    If I look at the original Macintosh Type 1 version of FF Thesis for example, then TheSans Plain (224 glyphs) has more than 3,400 kerning pairs. For those days (early 1990s) that level of detailing was quite unprecedented, but also demonstrated what could be done with digital type.

    While I agree that the goal should be to make as little kerning pairs as possible, I do think it is very helpful to start kerning the basic combinations, and go through all of them manually. It takes a while, but you will gain a lot of insight. With every new typeface after that you will recognise easier which combinations might need kerning, or (even better!) which shapes can be drawn differently to avoid kerning.
  • You might be right. So my numbers are with classes. 

    But I got some fonts from users that had even more pairs including classes. So you never know. 
  • Correct, Paul. Thank you for clarifying. 
  • AbiRasheedAbiRasheed Posts: 92
    Noted! thanks for the numbers Dyana and Paul. I think I'm bordering on that many pairs right now which likely means I need to work on spacing. Cheers
  • AbiRasheedAbiRasheed Posts: 92
    I just had something else to clarify since you guys mentioned spacing was important. If spacing is good, don't you still have to go through checking it manually pair by pair to make sure it looks good?
  • I wouldn't go through it pair by pair but look a texts. If I see a white dark spot I check if it is actually a spacing problem. And if it's not, then I add a kerning pair. 
  • Simon CozensSimon Cozens Posts: 156
    This may be a stupid question but how do you distinguish a spacing problem from a kerning problem? Is it just that a spacing problem shows up against key letters?
  • Kent LewKent Lew Posts: 488
    If you find yourself wanting to add a kern pair for a letter and a majority of other letters, then the sidebearing of the first letter is probably not optimum. That’s a spacing problem.
  • AbiRasheedAbiRasheed Posts: 92
    @Georg Seifert ah ok 
  • Chris LozosChris Lozos Posts: 864
    edited July 2016
    Perhaps Dyana is trying to teach AbiRasheed to fish instead of just to have food for one day? I think there would be great value in banging your head against many problem pairs to begin to understand how they duel with each-other.  This can teach you what a wonderful thing class kerning can be but also what pitfalls there are and what compromises a designer has to make to decide on the final kern selections.  By all means, don't rush through the learning process.

  • To test if it is a spacing problem you need to do a few things. First put the glyph that you think is the problem between some neutral glyphs (nnnxnnnooooxoooo). The do the same with the letters around it. Becaus if you think a pair has to much spache, maybe the pairs around it are to tight. 
  • I don't find kerning a laborious job at all. It usually takes me just a couple of hours to create the approximately 350 kerning pairs for each font.
  • edited August 2016
    It depends on how extensive you want to do it. Our next release have roughly 3500 kerning pairs per font (expanded, somewhere in the vicinity of 100 000 pairs per font). So, if your job is a couple of hours – imagine ours.
  • It depends on the design of the typeface as well. A more regularized rectangular face, or a condensed face for example you'll have glyph shapes that harmonize more naturally and require less kerning pairs.

    Like Frode said, how extensive your kerning pair lists are is also a big factor. I've settled on a default list somewhat recently for each master of lower to lower, upper to lower (includes mirrored lower to upper),  upper to upper, upper to appropriate numerals, lower to appropriate numerals, upper to punctuation, lower to punctuation, upper/lower/numerals to enclosers, numerals to numerals, punctuation to punctuation, numerator/denominator to fraction, and all diacritic collisions. The above naturally includes any case sensitive forms, alternates, lining figures, old style figures, and small caps.
  • Exactly. It depends on the design of letters. I experienced that, if there are too many anatomical inconsistency between the characters in the same shape group the spacing itself is not enough. It needs too many kerning combination of letters. It mostly happens on decorative and display typefaces. 
Sign In or Register to comment.