Generating weights - I started designing a regular (FontLab)

Hi there,

as a beginner in type design I am super happy about the great feedback and advice I've gotten on the forums here. I have ran into yet another problem about workflow and conceptual design of a typeface family.

Initially, I have started designing my typeface's regular weight, but eventually would like to create light and bold versions. For any kind of different interpolation (Multiple Master in Fontlab, or some Interpolation tools or scripts) you seem to need the extremes, which is rather logical.

However, I've started with my regular weight. Designing the light and bold from it seems a reasonably graspable task. My question though is this: How can I use interpolation between the extremes, but also not lose my precise work I have done on the regular? Is there a way to define a third state along one Multiple Master axis that solidifies a particular transitional point, i.e. my regular between the light and bold? Or is are my worries unfounded, and I will be able to find an "accurate" enough state resembling my regular between my extremes, which will be both designed based on the regular?

What's your approach to this, what are experiences in a similar situation? Do designers usually start off right away with light and bold only? Also, any advice in general about creating additional weights is appreciated.



  • Most designers, no matter how experienced, still design the regular first, simply because it is the most important weight. If your light and bold are designed decently, you can just create a MM file between the (extra)light and regular to create the lighter weights, and another between the regular and (extra)bold to create the darker weights. 

    I usually make the regular first, then the black or extra bold, then I extrapolate to create the light (and sometimes extralight) at a logical weight (depending on the difference in weight between your other weights), and then I clean up the extrapolated files. Extrapolation gets ugly fast, so 'cleaning up' is actually a lot of work.
  • Nick ShinnNick Shinn Posts: 1,641
    I make the Regular first.
    It embodies and clarifies the ethos of the typeface best, in particular stroke contrast, angle of stress and cadence (the new word for rhythm/colour—Erik Spiekermann and other cyclists should approve.)

    Then I make an Extra Bold and a Hairline, and interpolate these with the Regular.
    The hairline is made from a single path, to which “Contour > Make Parallel Path” is applied in FontLab.

    Finally, I make an Ultra Bold, by adding weight to the Extra Bold.
    I find that the idiosyncrasies of ultra-heavy weight produce problem interpolations, but, at the other end of the spectrum, the Thin is good for interpolation. 

    This method works for sans, and also for serif, but that can become more complicated.
    And as Jasper says, cleaning up of interpolations is necessary, especially if one is not rigorous about making equivalent BCPs in the master weights.

    For me, an important aspect of interpolation is picking the exact spots on the axis for particular weights.

    Then, adjusting horizontal stem widths and tweaking cadence—these qualities I consider to be “cleaning up”, and not something that can be built into an interpolation axis, because they do not behave in an incremental manner, in relation to weight.
  • Thanks for your answers.

    So Jasper, what you saying is it might not be too uncommon to use MM but have one from light to regular, and another from regular to bold? As opposed to designing light and bold (on the basis of a regular) and then interpolating everything in between?

    Nick, how do you technically set it all up? At this point I am most concerned with how to be flexible but still really make the most of using the regular I made. Are you using MM in your approach for the interpolation steps, or do you use something else, if I may ask? Did I understand you extract some weights, then leave the MM "basis" be and work on detailing those extracted weights separately?

    Curious to hear more, it really helps a lot figuring out a good approach.
  • Nick ShinnNick Shinn Posts: 1,641
    After I have drawn the Regular, I use this as a background layer over which I draw the skeleton path of the Hairline weight, with the pen tool in FontLab.

    For the Extra Bold, I start by beefing up the Regular with “Actions > Effects > Bold”, but this is a pretty rough starting point.

    I use the Blend function in FontLab.

    The detailing of the extrapolated weights has little or no bearing on the shape of the masters.
  • Chris LozosChris Lozos Posts: 1,302
    Unless the font is pretty geometric, I find it difficult to interpolate between a thin and an xtrabold italic.  I end up not using a master more than twice, then I interpolate between the corrected instances as masters.  I also have a tendency not to trust anything that seems automatic from software.  Doing the drawing and carefully correcting each instance is a requirement.  Anything that is too easy is likely to be wrong.
  • Make friends with the "Interpolate Node" tool in FL. It will save you a lot of time when drawing.
  • I find that you can generally do linear interpolation between the very thinnest weight you have, up to a bold (or thereabouts). Then there may need to be some discontinuity from there to the heaviest weight.
  • Chris LozosChris Lozos Posts: 1,302
    I agree, Thomas.
  • Thank you all for sharing. So from what I understand from your approaches is that MM is at best used to create some rough interpolations between manually created extremes and the already regular, which then get manually tweaked. Or alternatively just "blend" two extremes, without use of MM altogether.

    How about metrics in all this. Presumably you do metrics and kerning for the regular, then the extremes, so that the interpolation already produces something close to usable, no?

    Sorry about all the questions - maybe some things are obvious for veterans and long time type enthusiasts, but as a beginner there's a lot of process related things I am unsure about. So glad for your input.
  • Chris LozosChris Lozos Posts: 1,302
    "the interpolation already produces something close to usable, no? "
    Yes but don't rely on it straight as it comes, you still have to make adjustments.
Sign In or Register to comment.