I wasted two days recently trying to make a variable font using free software. The only ressource I found is a youtube video telling me to use a python script from a google drive.
It didn't work despite many many tries. I have two woffs of the same font that I bought and I wanna make a variable woff2 with only these two variants.
I couldn't believe that there isn't a clear way to do this.
You can use the Python library fontTools, which comes with a command line tool called `fonttools varLib`. Prerequisites: 1. you have a so called Designspace file which tells the tool where in the design space the fonts are and 2. The fonts have compatible glyph outlines and font features.
For the former, get a random font project like https://gitlab.gnome.org/GNOME/cantarell-fonts/-/blob/master..., install the Python tool fontmake in a venv, and run `fontmake -m Cantarell.desigbspace -o ttf-interpolatable`. In a new subdirectory, you'll find a new Designspace file and TTFs. Switch them out and edit the DS file to make sense. Good luck.
For the latter, you are probably out of luck because static fonts are usually compiled in an outline-incompatible way and your only options are to fix them yourself in fontforge or such or ask the original vendor, who is probably going to laugh at you and demand money.
there is a well-known piece of free software for this called "metafont" (also the original name for the concept here being neologized "variable font"—a "metafont" is an infinite set of fonts produced by an algorithm from input variables) but it doesn't have a gui and can't import woff or truetype
a variable font is a more generic term than a metafont. it’s a bit pedantic but while both have parameters, a metafont is usually parametrically generated based on its concept of a pen, whereas a variable font usually has different parameters that have been defined by manual bezier drawings. I don’t remember if metafont itself can produce a variable font but people have used metapost to go to SVG and then UFO (font source) from there
I think you may be missing the point. Metafonts have absolutely nothing to do with "variable fonts". It's a completely different branch of font technology. You cannot produce a variable font as OP intends using metafont.
I'm torn. Variable fonts are neat, but they also lay bare how woefully complicated every single layer of our software stack is where even displaying a single glyph requires interpreting a Turing-complete programming language. As I get older I increasingly just want a more transparent and understandable system that does less, even if that means giving up bells and whistles like automagic font kerning.
I'm a big fan. Variable fonts aren't just neat -- they're a godsend for graphic design when making things like posters, packaging, etc. To get the font weight just right, to get the letter widths just right. When dealing with a font that comes in just 2 weights or even 5, it's so common to want something exactly in between. Or when you want the typeface to be condensed, but the condensed version is way too condensed -- you want "halfway condensed" which is almost never a thing.
And I don't think most variable fonts require any kind of Turing-complete programming language? They're basically just interpolation. And interpolation is not that much added complexity for the rendering engine.
I agree with you that the kind of Turing-complete calculations that fonts these days are able to do, make me feel uneasy as well. But for me, variable fonts don't fall into that category. I look forward to the day when variable fonts are the norm for typefaces that come in different weights and widths.
I'd argue that it's unavoidable. These glyphs are meant for communicating to humans. Trying to reduce it removes the human element out of it, which can remove concepts like intent and context.
I think variable fonts are really neat—but I'm curious to know what professional typographers think about them. Take Concourse [1] by Matthew Butterick for example—if you click on the "PDF" link at the top, you'll get a PDF sample of the Concourse font, and on page 7 Butterick talks about how the font has been carefully duplexed so that different weights always take up the same amount of space. With a variable-width font, presumably you'd be able to adjust the tracking as well as the weight to make them match, but maybe not? And maybe not automatically? I dunno.
Do any professional typographers have opinions on the pros/cons of variable fonts?
H&FJ's (as it was at the time) Mercury was designed to keep the same spacing in different weights to handle the physical needs of printing a newspaper in different locations and conditions, see the design notes on the typeface [1]. That's the first time I remember this quality being touted in a font (among their set published weights, you couldn't vary it minutely, say, to get a bold-semibold).
While you can try to adjust tracking to make one weight of a face match the spacing of another, the faces are probably individually adjusted to make their own default spacing coherent within the face. It may be different enough between just two faces to sometimes not work, and even less across the entire range.
Adobe Acrobat used a version of Myriad and Minion as under-the-hood variable font substitution to make every PDF readable at least even if you didn't have the original typeface. The promise was you would never see Courier again. Multiple Master fonts made available this idea where you could generate specific typefaces along different axes for your design needs, but it wasn't a live thing, and you ended up with lots of discrete font files. My recollection of Multiple Master was that everyone hated them and were glad to see them die. Maybe the automated adjusting of parameters makes the implementation palatable now.
Great website but font prices are out of hand! Found one that looks awesome (West) but it's 350EUR if you want all styles! If you want Regular+Medium+Bold, that's still 150EUR. I bought Berkeley Mono back then for $75 and that is like the maximum what I would pay for a font. I am not a commercial user btw.
It takes a lot of time and skill to make a good font. Piracy is rampant and tricky to track down (let alone go after) people and businesses who use them without the correct licensing.
I originally submitted this article because I saw the NYT (maybe?) started using them. If you can view this: https://www.nytimes.com/interactive/2024/01/28/world/asia/ch..., I had to squint and saw that, "yes! that's a different 'F', and that's a different 'L'". It made so much sense, I thought "surely they didn't hire an artist to handwrite these captions. They must be using some kind of 'variable font'", and I googled just that and this site popped up. :) But then, maybe that's just an artist that works on these sorts of presentations.
You mean the handwritten text, “FINISHED SCHOOL” and such?
If so, even if you did that with text, you’d be very unlikely to use variable fonts at all, but rather just draw multiple versions of each character, and choose randomly or based on context (there are a few techniques). That’s waaaaaaaay easier than going variable would be—and variable still wouldn’t be sufficient to achieve the result on its own, though it could be a part of it.
But in this specific case: it’s in an image, the variations look unrealistically authentic for a font, and the framing lines and tiny illustrations and text are all done together. Pretty sure it’s authentic digital hand-drawn art. Won’t take long, genuinely just a few minutes for the actual drawing. The only hard part is fitting fluid-sized text content into the fixed-size illustration frames; that’s why at larger sizes it’ll be oddly spacious in places (and always inconsistently so), and since they’ve only taken care of it down to viewport widths of about 400px, below that (or if you use different fonts, or for a number of other possible causes) you start getting bad collisions.
New York Times decide it’s important to do things like this sometimes, so they get more people involved—not just a writer, but an illustrator and a developer.
Readit News
https://v-fonts.com/fonts/coronafaceimpact
It didn't work despite many many tries. I have two woffs of the same font that I bought and I wanna make a variable woff2 with only these two variants.
I couldn't believe that there isn't a clear way to do this.
Here's the youtube video: https://youtu.be/xoQuWARCUWI
For the former, get a random font project like https://gitlab.gnome.org/GNOME/cantarell-fonts/-/blob/master..., install the Python tool fontmake in a venv, and run `fontmake -m Cantarell.desigbspace -o ttf-interpolatable`. In a new subdirectory, you'll find a new Designspace file and TTFs. Switch them out and edit the DS file to make sense. Good luck.
For the latter, you are probably out of luck because static fonts are usually compiled in an outline-incompatible way and your only options are to fix them yourself in fontforge or such or ask the original vendor, who is probably going to laugh at you and demand money.
And I don't think most variable fonts require any kind of Turing-complete programming language? They're basically just interpolation. And interpolation is not that much added complexity for the rendering engine.
I agree with you that the kind of Turing-complete calculations that fonts these days are able to do, make me feel uneasy as well. But for me, variable fonts don't fall into that category. I look forward to the day when variable fonts are the norm for typefaces that come in different weights and widths.
Do any professional typographers have opinions on the pros/cons of variable fonts?
[1]: https://mbtype.com/fonts/concourse/sample.html
While you can try to adjust tracking to make one weight of a face match the spacing of another, the faces are probably individually adjusted to make their own default spacing coherent within the face. It may be different enough between just two faces to sometimes not work, and even less across the entire range.
Adobe Acrobat used a version of Myriad and Minion as under-the-hood variable font substitution to make every PDF readable at least even if you didn't have the original typeface. The promise was you would never see Courier again. Multiple Master fonts made available this idea where you could generate specific typefaces along different axes for your design needs, but it wasn't a live thing, and you ended up with lots of discrete font files. My recollection of Multiple Master was that everyone hated them and were glad to see them die. Maybe the automated adjusting of parameters makes the implementation palatable now.
[1]: https://www.typography.com/fonts/mercury-text/design-notes
Dead Comment
I dutifully clicked "Load 30 more" 12 times to see them all, but it'd be really helpful to be able to search and filter, and maybe "Load all".
If so, even if you did that with text, you’d be very unlikely to use variable fonts at all, but rather just draw multiple versions of each character, and choose randomly or based on context (there are a few techniques). That’s waaaaaaaay easier than going variable would be—and variable still wouldn’t be sufficient to achieve the result on its own, though it could be a part of it.
But in this specific case: it’s in an image, the variations look unrealistically authentic for a font, and the framing lines and tiny illustrations and text are all done together. Pretty sure it’s authentic digital hand-drawn art. Won’t take long, genuinely just a few minutes for the actual drawing. The only hard part is fitting fluid-sized text content into the fixed-size illustration frames; that’s why at larger sizes it’ll be oddly spacious in places (and always inconsistently so), and since they’ve only taken care of it down to viewport widths of about 400px, below that (or if you use different fonts, or for a number of other possible causes) you start getting bad collisions.
New York Times decide it’s important to do things like this sometimes, so they get more people involved—not just a writer, but an illustrator and a developer.
https://tug.org/TUGboat/Articles/tb24-2/tb77adams.pdf