How to be productive drawing type?

Hi, I'm new to type design and I feel the need for a process/plan to be more productive.
I don't make much daily progress as I find myself just keep perfecting small things rather than aiming for the finish line.
Also I'm not sure what's my finish line, goals or milestones.
The part I'm struggled with are punctuation, symbols, accented glyph, and italics.

Here's how I work currently:
  1. Lowercase and uppercase in Regular weight
  2. Spacing and basic kerning with classes for common pairs
  3. Create thin weight master, then interpolate for black master
  4. Punctuation and numbers
  5. Diacritics marks and accented glyph
  6. Italics
  7. More glyphs
What's your process/plan to be productive and get a typeface to finish?
What should be done on daily basis?
The finish might be in steps like version 0.1, 0.2 etc.
I think it's better to release smaller and frequent like those in Future Fonts.

Comments

  • Craig EliasonCraig Eliason Posts: 1,162
    One tip I would share is to not spend too much time perfecting any particular glyph. Move on to some other glyph or task. When you come back to it, problems that you couldn't see before will almost magically become visible. This can be true of a typeface as a whole. Setting something aside and coming back to it works really well. If you dwell on something for too long of a stretch, paradoxically it becomes more difficult to see critically.
    Interesting, because I've found myself kicking myself for not being careful enough with a glyph design, moving on too quickly and then realizing I've multiplied the poor decision across glyphs and even masters, and now instead of fixing one glyph I have to fix almost all of them! 

    On the other other hand, I find oftentimes it's only through that multiplication that the problem (and if lucky, its solution) becomes apparent. 
  • Completely agree with Mark. Even though I probably have less than a 10th of the experience he does, I've also found that my workflow for each typeface changes and improves based on the previous. Certain steps are interchangeable—there's no reason for italics to be designed so late into the process, for instance, and you can draw your figures and punctuation and diacritics before tackling the weights, if you felt like it.
    I have found it wise to be meticulous on metrics and very light on kerning (if touching that at all) until the very end.
    I do get better results when drawing a light and black master simultaneously, adjusting these to get interpolations that create a pleasing regular weight, compared to extrapolating a black weight from a light+regular. I've tried quite a few methods, including simply drawing three weight masters from scratch, but this one is definitely the most efficient.
    The only downside is that you can only go so far with your light and black master: an ultra black, for instance, needs adjustments that will mess up the regular interpolation.
    @Craig Eliason This is why I love elements in FontLab (I'm sure other font editors have similar functions). Change one serif, stem, or shoulder, and every matching serif, stem, and shoulder, will do the same. This is also very useful in order to judge the typeface as a whole, within a sentence or paragraph.
    And on this note, I think that making good use of all the workflow improving features that your font editor has to offer has just as much to offer your efficiency as getting your steps in the right order. Through trial and error, I've learned so much about using FontLab that is improving both my work speed and quality.
  • Indeed there's nothing more effective than looking away for a while. Which might justify rushing in proportion to earliness.
  • Chris LozosChris Lozos Posts: 1,371
    There is no easy path. I start with the glyphs that made me want to design the face to begin with so I can see the character of the face. Then add n, u, o, a, v, d, and s. Spend a lot of time with test words before you make too many letters.  Let them fight with each other until you have solved most fitting problems. Be sure to then add your most dreaded problem children and have a good skirmish with them. Put it away for at least a week and see what needs fixing.  Percolation time is very important. Absence makes the eyes grow keener.
  • Ray LarabieRay Larabie Posts: 1,108
    edited May 24
    My diacriticals got easier when I learned to love them. Doing the accents used to be a nuisance but no longer. The more you learn about the possibilities, the less they feel like a chore. I'll admit that mathmatical symbols and ordinals are still a drag.
  • Chris LozosChris Lozos Posts: 1,371
    I have been thinking about doing my display fonts with math symbols from one of my other completed fonts.  They are there if needed but less work other than minor height adjustment.
  • @Mark Simonson Is it okay to not draw glyph you'll know it's likely won't be used? Maybe draw them later by request?
  • @Matthijs Herzberg I also use Fontlab, and I think I should utilize elements more than what I currently do. Usually I only use it for serifs and bowl. There's more to learn from the tools.

    I used to draw Thin and Black masters, the reason I start with Regular Master is I can judge whether this will look good on regular text usage first. I also found the interpolated regular doesn't look great, but that's because I didn't know about Intermediate Master layer (:wt=400).

    I think I need to watch more type drawing process, if there's any, since the process usually takes a long time and not many document their process in details.
  • Maybe draw them later by request?
    That saves time and can make more money, but unless your font is pretty unique a potential customer who notices missing stuff will usually look elsewhere (and you'll never even find out).
  • Mark SimonsonMark Simonson Posts: 1,351
    @Laurensius Maybe, but my understanding is that, at least for some larger potential customers, it may not pass some technical requirements for font purchases if expected standard characters are missing.

    My advice is to spend as little time on them as possible unless you think they may potentially be used (e.g., for a text face).
  • I would recommend not to tackle kerning until the final stages. Create classes, sure, even detailed classes, but do just spacing with a lot of care (it’s a part of design) and save the kerning for the final stages.
    Interesting, because I've found myself kicking myself for not being careful enough with a glyph design, moving on too quickly and then realizing I've multiplied the poor decision across glyphs and even masters, and now instead of fixing one glyph I have to fix almost all of them! 
    I think @Mark Simonson was thinking about losing oneself in details. Clearly the important things can benefit from being adjusted and perfectioned first (stem and contrast consistency, punctuation, etc.). What’s important, in my experience, is to take a break often, especially if you have the tendency to over-refine curves or non-essential details and things.
  • Alex VisiAlex Visi Posts: 132
    edited May 24
    I separate design and production (things which don’t involve any new decisions), and try to stay away from production for as long as possible. The reason is simple — as you add stuff, things you have to maintain grow exponentially and so grow repetitive tasks (which of course lead to wasting time, forgetting, being lazy and so on).

    In other words, I start by making the font with minimum charset, no kerning etc, as good as possible, test it, use it, see how it works, from which you can see problems to fix. And only then continue to extending it for other languages, adding extra symbols.

    Working on a few styles in parallel seem to be an essential part of that design process, but it depends.
  • mathematical symbols and ordinals are still a drag
    It's a double drag. Not only are they boring to make, you know that in most fonts (especially display fonts) most of them will never be used. 
    I actually like to draw them. But since, as you say, if a typeface is not studied specifically for Math with an extended math symbols set, they are unlikely to be used widely (except the main ones), and since they benefit from standing out, I am experimenting to have general sets which can be used across width and weight styles.James Puckett said:
    Here’s how I work:
    1. Kern with MetricsMachine. Proof kerning on paper. Fix problems in Glyphs. Repeat with subsequent masters
    @James Puckett: May I ask if by kerning with MetricsMachine allows you to go "back and forth" with Glyphs (or Fontlab).
    So far I preferred an approach where I am building text files with words that actually make sense and/or are more used in the more spoken languages, I find hard to kern out of context or single pairs on their own, I also think it’s tricky. 
  • Matthijs HerzbergMatthijs Herzberg Posts: 114
    edited May 24
    @Claudio Piccinini
    Only kerning words in the more spoken languages is not a fool-proof approach for a simple reason: there will be tons of combinations in less-spoken languages that you're bound to miss (assuming that you get every combo in English/Spanish/Etc, which you probably also won't).
    It's really best to go over a kerning string with every combo (aaabacadaeaf...), for both uppercase-uppercase, lowercase-lowercase, and uppercase-lowercase. This seems tedious, but if you're properly using class-kerning, it becomes much easier.
    If you have a hard time judging how much spacing you need, add an "nnn" or "HHH" or "non" or whatever helps you see the desired spacing clearly between each kerning pair.
    Or make a row of "nnoonnoonnoo...", then a row of kerning pairs, repeat.
    Good thread on kerning here.
  • James PuckettJames Puckett Posts: 1,785
    @James Puckett: May I ask if by kerning with MetricsMachine allows you to go "back and forth" with Glyphs (or Fontlab).
    So far I preferred an approach where I am building text files with words that actually make sense and/or are more used in the more spoken languages, I find hard to kern out of context or single pairs on their own, I also think it’s tricky. 
    I kern everything in MetricsMachine first, import the kerning to Glyphs (File>Import>Metrics), and then revise individual pairs in Glyphs.

    Don’t try to kern with words in common languages. You will miss hundreds of pairs that you don’t think are common. American designers need to be able to use the names of towns in Slavic countries and those have pairs you would never imagine. And you need to learn to see pairs in generic contexts or else you’ll end up with pairs that work right in the word you tested but don’t work right in many other words. Get started with my collection of MetricsMachine contexts.

  • Next question then: How's your to-do list, daily/weekly productivity system looks like?

  • Chris LozosChris Lozos Posts: 1,371
    I don't have one, I just do what I do.
  • Mark SimonsonMark Simonson Posts: 1,351
    I tend to be a sprinter, working in concentrated, focussed stretches contrasting with periods when I'm doing things other than working on fonts. Part of the reason I'm able get away with this is that most of my font work is on retail fonts where there's no prescribed deadline. I don't do commissioned fonts nowadays, but when I did I had to be more disciplined about it. In those cases, deadlines and milestones drove my schedule.

    That said, I do have a tentative list of the next few fonts I want to finish over the next year. I usually have three faces in the works and switch between them. 

    A significant part of my time goes to doing updates and fixes for released fonts. This is almost always driven by customer requests and feedback, including major things like adding cyrillic, additional styles or features. Updates tend to take priority over new releases.

    I have a tendency to procrastinate, but I've found that if I just start doing something—anything—on a font, that will be enough to get things moving.

    I don't know whether any of this would work for someone else, but it's a "system" that works for me.
  • Just a note for @Matthijs Herzberg: I do not mean to say you’ll produce “foolproof kerning" by making such selection. It’s just a series of files I’m trying to assemble to have a good base in a preliminary final stage of a typeface.
    But I still do not find the "every combo" (aaabacadaeaf...) approach effective, not because it’s tedious but because it does not show you the pairs in real context.
    I think I have read and re-read most of the threads on kerning here, and many resources.

    @James Puckett: Of course, I mean to look at pairs in more contexts, as I said I am just trying to create selections where 1) actual words are represented, and choosing the more common to start with; 2) have a varied but limited set of words with more critical pairs.
    How’s the file you posted to be used? I do not own Robofont and thus no Metrics Machine, I guess.
  • James PuckettJames Puckett Posts: 1,785

    How’s the file you posted to be used? I do not own Robofont and thus no Metrics Machine, I guess.

    MetricsMachine displays pairs in contexts. For example, HH<PAIR>DD<PAIR>OO. You can create as many contexts as you want. I have contexts for uppercase, lowercase, numbers, different types of punctuation, and math symbols.

  • How’s the file you posted to be used? I do not own Robofont and thus no Metrics Machine, I guess.

    MetricsMachine displays pairs in contexts. For example, HH<PAIR>DD<PAIR>OO. You can create as many contexts as you want. I have contexts for uppercase, lowercase, numbers, different types of punctuation, and math symbols.
    Ah, I see. So you need metrics machine to use your files, right?
    At any rate, I miss the old Fontstudio feature, it was SO useful. There are online resources to generate texts, but none has that immediacy and ease of use.
    "But I still do not find the "every combo" (aaabacadaeaf...) approach effective, "
    It works for me very well but I also follow it with many pages of real text in languages covered. I find I need both.
    Yes, I was trying to strike a balance between "wall of texts" including pairs which will never present themselves as critical and actual use. I will probably do the same, in the end.
  • Chris LozosChris Lozos Posts: 1,371
    @Claudio Piccinini
    " Yes, I was trying to strike a balance between "wall of texts" including pairs which will never present themselves as critical and actual use."
    It may take a while to set up a sequence of text files that first take care of common class kerning followed by class to individual kerning, followed by the rare individual to individual kerning. Once you set up your logical system, you can go through it in 2 days per font [typical 900 glyphs]. With experience, you learn which are the typical pairs in a given language which can cause problems so you put together paragraphs with the problem children per language in them.  The brute force approach minimizes the "big surprise" pairs and allows you to ignore the ones which never occur.  I also used to use FontStudio years ago and fondly remember how it worked ;-)


  • @Claudio Piccinini
    " Yes, I was trying to strike a balance between "wall of texts" including pairs which will never present themselves as critical and actual use."
    It may take a while to set up a sequence of text files that first take care of common class kerning followed by class to individual kerning, followed by the rare individual to individual kerning. Once you set up your logical system, you can go through it in 2 days per font [typical 900 glyphs]. With experience, you learn which are the typical pairs in a given language which can cause problems so you put together paragraphs with the problem children per language in them.  The brute force approach minimizes the "big surprise" pairs and allows you to ignore the ones which never occur.  I also used to use FontStudio years ago and fondly remember how it worked ;-)
    Yes, that would be the goal. Do you know specific text files which are best for the "brute force approach"? I have collected some but I am still not so satisfied.
  • Chris LozosChris Lozos Posts: 1,371
    I use the same files as many.  Just search and replace until you are nauseous ;-)
Sign In or Register to comment.