Rounded Stem Values / Kerning Values

This goes for Masters but more so for interpolated styles/weights. Is anyone else obsessed with having even number stem and sidebearing values? (e.g. 30pt vs. 31 or 33pt) I know it's not always possible/desirable, being that in hairline weights the vertical/horizontal stems usually have a difference of only 1pt. (e.g. Hor=4pt Ver=5pt)

Further, I know many kern in incremental values of 5pts, I imagine some do as well in values of 2pts, but does anyone further use a Round Transformation on interpolated styles/weights in MetricsMachine to clean up kerning values and ensure they round to values of 2, or 5pt?


  • Yes, I go you one further my SBs or factors of 2 but my kerns are factors of 4.
  • It makes me think there should be a way to force-round the values of generated instances from Superpolator. Is there?
  • There’s a way to select the interpolation preview in Superpolator and it will show you stem widths, but I don’t remember exactly how. The problem is that between caps, lowercase, and figures something usually ends up with odd numbered stems. But you can always just interpolate different glyph types separately and combine them into fonts. 

    Rounding stem values would require a tool that measures individual stems and adjusts interpolation values accordingly. 
  • But why would you want to do that ?
  • > I know many kern in incremental values of 5pts, I imagine some do as well in values of 2pts,

    I do with increments of 1unit or more, personally. As long as identical sidebearings have the same kerning value, the rest looks like arbitrary restrictions to me unless I'm missing something.
  • Michael JarboeMichael Jarboe Posts: 207
    edited February 2015
    Jacques, Adrien, not really sure. I just wanted to get a sense if other designers have any kind of particular process/logic pertaining to all these various values. Trying to decide myself if there's any benefit to adhering to any particular set of rules moving forward.

    It just looks 'cleaner' and seems more disciplined when seeing kerning/stem/sidebearing values that are even number values and/or rounded values. Functionality-wise it's pretty trivial, and process-wise any of this is maybe a bit obsessive-compulsive.
  • I agree that functionality-wise it's trivial, but it is also negative functionality. That is, you are restricting yourself for no particular reason. Although I will start at nice even values, I will end up with all sorts of other values by the time I'm done, after tweaks and such. Plus, as noted, at extremely light weights you have no (reasonable) choice.
  • Depending on what terms you prefer, and I use "rounding" for the act of locating control points and set widths on pixel boundaries before rendering, there was quite a bit of discussion in the past on the benefits of what we called "quantizing", if you want to search on that.  And "pt" is a semi-good term to use in type sizing, but unit, as in unit per em, is more important in the type design process. 

    In summary, assuming it doesn't interfere with the perfection of a design in any direction, black or thin, simple or complex, I think the benefits of quantizing contours include; mathematical simplicity during analysis and design, sizability advantages for file reduction, advantages related to font sizing at low resolutions, and those rare synch-ability issues related to frames per second when type is being animated, the later of which, is a distant offsprung of Superpolation's rounding issues. 

    I agree with the apparent consensus that there are two widely separate camps on this. Camp A has the kings and queens of high resolution, who don't rightly care about the relationship between input, (in units per em), and output (hopefully in 1,000s of pixels per inch)... pretty much ever, and Camp B where some level of obsession with the relationship between input and output exists because there is less than 300 pixels per inch in the device(s) involved in displaying the type most of the time, and/or the type is small much of the time.

    An then, in a middle, Camp C, are those just arriving or moving from one camp to the other with their possessions. If you meet a tool-maker, make sure they're in Camp B, as you can use Camp B tools perfectly well to make Camp A fonts.:)
  • and then there are some of us who are madly in love with the progression of 8 bit numbers ;-)
Sign In or Register to comment.