Autotracing is Boring and Autotracing is Exciting

Ray Larabie

@Hrant H. Papazian
Whachutalkingabout? Autotracing is exciting as hell. Perhaps you might consider straight up autotracing of historical fonts boring and I agree, that could be boring. But it's kinda neato sometimes.

I find autotracing very useful for producing special effects. If I need to produce a layer of newspaper halftone or stipple, it's pretty much impossible to accomplish without autotracing. One of my typefaces called Mochon which was drawn with grease pencil would be been impossible to make without autotracing. If I had manually traced all those shapes, I don't know if the boringness level would be affected.

I went through a phase almost 15 years ago of using existing metal/phototype fonts, autotracing, manually cleaning up, adding special effects, autotracing and manually cleaning up again. At the time I thought of it like sampling music and creating something new. I tried to make it clear in the descriptions that it was based on old material. I think some of the results are good but some of it doesn't sit well with me in 2021. At the time I was inspired by sampling in music, these days I'd avoid it...I cringe a bit when I see some of them. But I sure don't feel bored.

Teeshirt started as an autotrace of an old American Typewriter specimen. I manually cleaned it up, expanded the character set, and used that to create layers in Photoshop. I painted real teeshirt fabric with acrylic paint, cracked it and scanned it. It would have been about the same amount of work to make an original typeface and much less work to use one of my own fonts. But I couldn't. The whole point of the typeface was 1970's teeshirt and it needed an authentic 1970's typeface for it to work properly. I could barely justify it to myself at the time but my desire to make something cool overrode the dubious ethics of it. Nowadays, I wouldn't attempt it.

Tight...so embarrassing. At the time I was unable to track down Dean Morris to ask permission. A couple of years after I made it, he called me out on it, publicly. If any of you young designers are thinking of doing this sort of thing, consider that you might feel stupid about it later and the original designer will make you feel like a dick.

Isn't that just a little bit interesting? C'mon, Hrant. All the cool kids are doing it.

Simon Cozens

Ray Larabie said:

I find autotracing very useful for producing special effects. If I need to produce a layer of newspaper halftone or stipple, it's pretty much impossible to accomplish without autotracing. One of my typefaces called Mochon which was drawn with grease pencil would be been impossible to make without autotracing.

In general this is probably true. But I feel compelled to point out some of the awesome things that Luke Prowse has been able to do through scripting, and that his scripts are open source. (I was involved in tidying up the Python but the logic and ingenuity is all his.) Autotrace for this kind of stuff makes big heavy files with loads of points, but doing it generatively through scripts makes quite neat and tidy font files at the end.

John Savard

Ray Larabie said:

I find autotracing very useful for producing special effects. If I need to produce a layer of newspaper halftone or stipple, it's pretty much impossible to accomplish without autotracing.

This of course immediately brought Roger Excoffon's Calypso to my mind.

Craig Eliason

@Ray Larabie Any thought of taking down those fonts that are embarrassing and made you feel like a dick? (That’s a sincere question not a rhetorical dig.)

John Savard

Craig Eliason said:

@Ray Larabie Any thought of taking down those fonts that are embarrassing and made you feel like a dick? (That’s a sincere question not a rhetorical dig.)

Of course, no doubt he will answer you soon, but my guess would be that he would have done so long ago had he been selling the fonts himself, but the typeface belongs to Typodermic now.

Daniel Benjamin Miller

What is truly boring is autotracing for text fonts. For jagged creations, of a more distinctive character, it works as a method. I could autotrace an image of $textfont pretty easily; that would be boring and would just produce a worse version of $textfont.

Ray Larabie

@Craig Eliason If they weren't potentially useful I might. The resulting fonts being actually good, tips the scale.

Nick Shinn

I can see it might be useful for the textured stuff Ray mentions, but for me, I’ve tried it a few times on solid types, and it took so long to clean up the autotracing, why bother, so I just manually trace over a scan. But I rarely use scans anyway.

Helmut Wollmersdorfer

Simon Cozens said:

 Autotrace for this kind of stuff makes big heavy files with loads of points, but doing it generatively through scripts makes quite neat and tidy font files at the end.

That's a big problem in case of fine details like initials. E. g. the Cloister Initials http://luc.devroye.org/fonts-88654.html are available in some free versions, obviously autotraced, and they take long to render in a website.

Helmut Wollmersdorfer

Autotracing is just a way to reconstruct fonts.

Reconstruction of historic books needs reconstruction of the original fonts. Existing reconstructions of the same typefaces are not usable, because they are "improved". Also some disappeared, got out of fashion.

Automatic reconstruction of fonts is a scientific field. It's more than just autotracing, recognize the features like e.g. serifs and apply them.

Automatic construction of fonts is important e. g. for Chinese. This is done by a few samples for training the style of the strokes and apply the style to skeletons of all characters.

One impressive example how computers can automatically learn different styles:

Learning a Manifold of Fonts

PabloImpallari

We started with autotracing from old ATF scans and then redrawing new shapes but keeping the autotracing shape in the background layer as reference for most of out libre series: libre caslon, libre baskerville, libre bodoni etc... but for libre franklin we did it a bit diferrent: we also drew new shapes, but we followed the ink very very closely to the the effect it makes on paper intead of a pure shape. For example in the letter /A or /G the is no way you can produce the nice friendly roundness of the old specimens if you end the stems in just a corner node. Also you cant if you use a stamdard geometrical roundess as in rounded font (like Dosis).

As a result each stem corner of each letter have a different degree of roundness that only applies to that particular letter.
(A V W G R Y Z C M N all different corners details. It was not lasiness of cleaning but a desing desition instead. In fact it was more work. Shap corners would have been more easy to do and far less nodes to work with)

And I have to say it: I ended up really really loving Libre Frankling and I must not say this since its my own font and can sound rude. But I thing os the best Frankling revical out there. Even better than commercial ones.. ok. Maybe not better, but friendlier to the eyes and closer in feeling to the originals from ATF.  And Libre Francking was a hit. A home run. And instant succes!

Commercials revivals feels to me too damm clean and too damm sharp and I hate that super cleaningness. Especially if they are going to be used for headlines h1 in the web than now a days is mostly fantastic phones with fantastic resolutions screens. So I also love the results of autotracing if you can use the results to study and understand effects of ink on paper, effects of gaussian filter of the scanner, effects of irregular metal letters. etc.

Basically autotracing is just another tool that helps you to see details. As all tools: Is not good or bad per se.. its what you use it for and how you use it.

Nick Shinn

Copying an old printed text sample, revival or restoration, needs to take account of the two media it bridges.

“The task of renovating or recreating a design from old impressions is the most difficult of all. The effect of impressing upon damp paper, of worn type, and of the spread of ink, have to be reckoned with; and great skill is needed if, while removing blurred outlines, the subtleties of the original engraving are not to be lost.”

—Stanley Morison, A Tally of Types

The gulf between metal and digital typography cannot be mechanically bridged, as
the transition to uniform shapes occurs from characters which vary in design from size
to size, and in quality from impression to impression. Any attempt to systematically average an ideal letterform leads to blandness. 

What if one searches for the idea or best instance of each character? Unfortunately, there is no such thing, they all need cleaning up. That will be done by eye, so why not proceed in such a manner from the outset? And rather than put faith in the seeming veracity of a photographic simulacrum as one’s model, trust to the authenticity of “still life” drawing.

Tracing over a scan is also quite an interpretive exercise.