Standardizing Curves When Digitizing Typefaces, "Drawing Letters"

noahtheodoresimon's picture

i am digitizing a full alphabet for the first time, to be used as part of an identity package. i had flirted with this kind of work in the past with individual characters, but always stopped short of a full alphabet.

discussions like this one have been extremely helpful in addressing general best practices for vector technique. however, they have largely failed to address the very specific considerations involved in drawing digital type.

my remaining concern is for the finest level of detail — the best approach to bracketing a joint, capping a serif, etc.

i've taken to inspecting H&FJ letterforms for reference, as they seem to be thorough in terms of optical sizes, consistent in their execution, i like them.

i'm interested to hear from the community about everyone's respective approach to achieving the best results on this purely technical level.

the examples I attached are details HTF Didot 96 Light. here's what i see as odd/notable/worth considering as I proceed, corresponding to numbered images:
1. serifs terminate in rounded caps, but not semi-circular caps as would a round-capped stroke in illustrator. the curvature of these caps is standardized. likewise, the curvature of the joints is consistent throughout — distance of anchor points from hypothetical corner, positioning of handles, everything. this leads me to infer that the responsible parties have developed a method to standardize these elements.
2. however, in some cases, curves I would expect to be consistent are not quite consistent, leading me to believe that the responsible parties are simply eyeballing it — extremely accurately — or that some judgment beyond my abilities has been applied to this case.
3. strokes, though they appear consistent throughout, are actually inconsistent between horizontal/vertical/diagonal instances. this formal inconsistency appears to be consistent in its application.

what is the best way to address these issues? the simplest aspect of the question seems to be that of standardization of brackets/caps (which seems like it should be an easy software issue, but i have found no easier way than transplanting one curve over and over. this is tedious; seems stupid; i don't want to do it).

the questions suggested be 2-3 seem to be broader questions of drawing methodology. i have inferred certain relatively common, deliberate inconsistencies in letterform design (the lower curve of a capital S, O, J, etc. descending below the baseline to achieve the appearance of sitting on the baseline, for example) which appear to be relatively standardized. are the issues of 2-3 comparable? if so, how does one best negotiate them?

many thanks.

1.jpg135.32 KB
2.jpg195.02 KB
3.png56.78 KB
oldnick's picture

1. "the responsible parties have developed a method to standardize these elements": yes.

2. "some judgment beyond my abilities has been applied to this case": yes.

3. "this formal inconsistency appears to be consistent in its application": yes--it has to do with a documented anomaly by which the human eye perceives horizontal and vertical dimensions differently.

noahtheodoresimon's picture

ok ok

1. what is the easiest way to standardize these kinds of elements?

blank's picture

what is the easiest way to standardize these kinds of elements?

Work it out in the prototype stage and copy/paste from there. Fontlab has find/replace functionality that can be applied to an entire font but it does not work especially well.

Nick Shinn's picture

…this purely technical level…

Type design is 100% "technical"—to do with technique—as Peter Biľak explains:

The element/level of design that you are considering here is the opposite of technical (in the general sense of something that may be reduced to standard procedure), and is entirely subjective and a matter of taste and discretion, which may only be developed/acquired through study and practice.

Rather than standards in executing such things as overshoot, there is the consensus of general practice, and exemplars—reliable practitioners.
But you will soon reach your ceiling as a designer if you only follow precedent, and don't experiment and make mistakes.

noahtheodoresimon's picture

Alright, thank you — I think these are very helpful points. In cases like overshoot, in which the consensus of general practice has rendered some approximate standard, such standards are useful to know (and I didn't know that one). As long as I'm not slaving away solving problems that have already been solved (within the bounds of reason), I'm happy to do it the slow and difficult way — which I agree is generally the most auspicious. Many thanks again.

Nick Shinn's picture

Here's an interesting piece of overshoot trivia: Univers Light has overshoot on the "I".
This was a feature of the face when it was first introduced (mentioned in advertising), and has made it into digital.

IMO, PostScript has resulted in typefaces becoming far more regularized than with previous technologies, constrained by Stem Width and Alignment Zones for hinting purposes. Before the invention of Alignment Zones, the concept of overshoot did not exist for the type designer as such a neatly visible, regularizing imperative.


I'm presently working on a rounded sans—it's a real pain to figure out the overshoots.

1985's picture

Nick, that's precisely the kind of feature that, had you not pointed out it's intended presence, I would have put down to poor quality digitising. I wonder why it overshoots at the top and not at the bottom too?

As a novice, examining existing fonts can create a bit of a dilemma. On the one hand a feature may exist a certain way because the designer possesses some judgement beyond your abilities (as you put it) This is a conscious decision on their behalf, but with only limited experience one can't necessarily fathom it. On the other hand the designer might indeed have made an error, or not yet be experienced enough to do a more sophisticated job (not likely with H&FJ, mind). When I begun I knew of fewer optical corrections than I do now. With that in mind I imagine there are more mysterious corrections to learn about such as the one that Nick has posted above, the more mysterious they get the less general consensus will surround them, I imagine.

Also bear in mind that fonts are drawn on an explicit grid (unlike Illustrator) and this may account for some (rather frustrating) discrepancies.

Syndicate content Syndicate content