Slightly differentiating a glyph’s instances

Domenico Statuto

LibreOffice has an extension named Patina which introduces slight differentiations among the various instances of the same glyph, just as once happened when using lead type in letterpress. I wonder if is there something like that for Indesign.

https://extensions.libreoffice.org/en/extensions/show/patina

J_Tillman

Read about Playpen at type-together.com as they seem to have a “shuffler” built into the font.

Thomas Phinney

You can code the font so as to do this using contextual OpenType code.

Unfortunately, many layout engines break the “context” at word boundaries, mostly for performance reasons.   

When Bickham Script first shipped, you could type “the the the the the the” in Bickham, in Adobe InDesign, and every “the” would be different. But later that was broken by Eric Muller’s changes to Adobe’s CoolType engine. Sigh.

Ray Larabie

@Thomas Phinney Won't it work if you use dummy alternate spaces? Let's say you're doing a 5 shuffle, you'd have space, space.2, space.3, space.4, space.5. That way it keeps on rolling, except at line breaks and non-breaking spaces.

Simon Cozens

The run would get broken into spaces at the character level (where there is only one space character), before you get to the space substitution at the glyph level.

Kent Lew

We applied a similar OpenType contextual code at FontBureau in the Magnet style of Cyrus Highsmith’s Icebox font. I worked with Cyrus on the “randomization” features.

Of course, the {calt} feature is not truly random. So for kicks, I also wrote a “Fridgerizer” applescript for InDesign that would further twiddle the setting by individually applying a random {salt} variant character by character. It would also randomly colorize the letters with a user-selectable set of basic colors.

I thought it would be fun to give this script away with each license. But FB didn’t have a good administrative mechanism for managing such a thing back in 2015. (And I wouldn’t have wanted to maintain this. It was just a fun gimmick.)

That said, this seems to be the basic approach of the OP’s linked LibreOffice extension — randomly applying some amount of baseline shift, rotation, and perhaps some slight scaling ( ? : my Italian is rusty) character by character through a setting.

This would certainly be possible to script in InDesign. But I don’t know if anyone has done it.

It’s hard to judge the value of such processing when the examples use Igino Marini’s Fell Types which already have a good amount of “patina” already. I question how effective it would be when applied to a more conventional digital font.

Seems more likely to end up looking like early LaserWriter printouts than incunabula or letterpress printing.

John Hudson

It isn’t clear to me where in the Il Coville Patina plugin for LibreOffice the randomisation is happening.

Produced in the sense of this output for a randomising font was made by the Patina plugin, or that the randomisation of the outlines was done by Patina?

Randomisation has taken two forms in digital typography: pseudo-random cycling of glyphs via contextual substitutions, and actual randomisation of outlines. The latter is much more rare: the best known example of such randomisation was the original Type 3 version of FF Beowulf, which used Postscript code to randomise the outline shapes in the print stream, so every time you printed a document set in the font it looked different.

Kent Lew

If you look closely at the examples in the image, the basic glyph contour, with its distinctive bumps, does not really change. What does seem to be varied is a combination of glyph width via scaling (with a resulting degree of weight variation), some slight baseline shift, and perhaps some slight rotation (as indicated, I think, by the screenshot of the LibreOffice character setting interface; but maybe not).

It’s not a cycling of any glyph variation in the fonts, and it’s certainly not any real randomization of contours. So perhaps the description might be more accurate as “These distortions of the Fell Types letter s are produced by Patina.”?

Nick Shinn

I’ve made several pseudo-random typefaces via OpenType feature, starting out imitating the “organic” variations of handwriting and hand lettering, then distress.
I never did get around to mimicking the vagaries of letterpress and lost interest in the process, but recently picked it up again.
This is my homage to Jamie Reid (he of the Sex Pistols branding), who died last year.
There are no doubt many more kinds of pseudo-randomization possible with OpenType substitution, waiting to be discovered.

Domenico Statuto

Thank you so much for your comments. It’s better to wait for the day such a feature will be natively included in either the font technology or in Indesign. Kent’s comment that the final outcome would probably be less attractive than hoped for, has put the nail in the coffin. Indeed IMFell fonts are not usual, so what does work with them doesn’t work with usual fonts. But I have Libreoffice, so I could try Patina with normal fonts.

Thomas Phinney

I should also point out that the OpenType format also defines a layout feature specifically for this purpose: 'rand' (randomization). It is not widely supported, however.

As long as you don’t have the same word a lot, or immediately repeated, the pseudo-random rotating glyph effects can be good. One might also look at how the OpenType version of LTR Beowolf works. The combination of three separate fonts PLUS the pseudo-random glyph rotation is nice in Beowolf, same concept would work well for something like the Fell types. In typesetting such a font family, if you have a second “the” in close proximity, or a sequence like “the theory” you can set one of the two in a separate font to get a completely different glyph set. It would be a workaround, but a pretty effective one. Whether one bothers with such extra steps would depend on available time and the nature and quantity of the text being set.

Obviously this doesn’t work with existing fonts (aside from Beowolf), so this is more advice for people making fonts at least as much as those using them, of course. One thing not to be lost here is that making 4x or 8x as many glyphs can be a heck of a lot of extra work. Either the variation is done manually (tons of work), or by script or automated process (potentially a bunch of work to set up). Not to mention that all this will make for some very large font files.

Nick Shinn

I made several typefaces with four versions of each character, but subsequently came to the conclusion that only two vertsions are necessary for pseudo-randomization. The reason is that characters rarely* repeat within the same saccadic fixation. With suitable coding, it’s possible to configure a two-version pseudo-randomization that doesn’t repeat a glyph until the fifth character—which is in a separate fixation, or at least outside the fovea’s range to discern with sufficient acuity to notice randomness. I doubt that readers use their “spot the difference” visual software across different fixations, because their primary task is to decode the text, and once that is done further visual recognition of glyph features, such as whether a glyph is repeated exactly, is redundant, and not pursued.

*Numerals being an exception.