I've always found this odd as well that when licensing web fonts, typically the foundries don't allow you to subset (or provide a subset version of their own).
Licensing fonts is just such a headache that I would never want to deal with it. They are usually based on metrics that are pretty cumbersome to track.
If you want a custom font where no OFL licensed alternative would work I'd rather hire someone to design a new font. I guess that is also the reason so many companies now have their own custom font (e.g. Spotify, Adobe, etc.).
Just beware of OFL with a Reserved Font Name, you can't modify them (and building from source counts as modification) and redistribute without renaming them. This is especially a problem for Debian since they like to build everything from source, including fonts.
In a world where art illustrations are being copied left and right by AI what is stopping the exact same thing to happen to typefaces? (And with it any license inconveniences)
The design of typefaces aren’t copyrighted in the U.S. The only thing that is protected is the software: TTFs, OTFs, etc. [1] That’s why so many clones of popular fonts (and old metal type) exist.
These days the value in a font isn’t in the letterforms, it’s in the kerning, ligatures, variability, etc. which all flows from the software. It’s also where a significant amount of the labor in creating a typeface comes from. And it’s the thing that sets apart professional-quality fonts from many (but not all!) free ones.
If AI can write new font software by cloning bitmaps of letterforms _and_ get the kerning, ligatures, variability, etc. right… it’ll change the type foundry industry in a big way.
I do absolutely think integrating into the deeper features and functionality of fonts and opentype is what will make it work! Making it work well with existing font design software is also super important for typedesigners.
I used to buy fonts, but font foundries made such a mess with licensing that I only use OFL licensed fonts now.
> Allow subsetting
I've always found this odd as well that when licensing web fonts, typically the foundries don't allow you to subset (or provide a subset version of their own).
If it's not OFL, it goes in the bin.
Licensing fonts is just such a headache that I would never want to deal with it. They are usually based on metrics that are pretty cumbersome to track.
If you want a custom font where no OFL licensed alternative would work I'd rather hire someone to design a new font. I guess that is also the reason so many companies now have their own custom font (e.g. Spotify, Adobe, etc.).
Just beware of OFL with a Reserved Font Name, you can't modify them (and building from source counts as modification) and redistribute without renaming them. This is especially a problem for Debian since they like to build everything from source, including fonts.
https://wiki.debian.org/Fonts/Bugs/rfn-violation https://wiki.debian.org/AutoGeneratedFiles
In a world where art illustrations are being copied left and right by AI what is stopping the exact same thing to happen to typefaces? (And with it any license inconveniences)
The design of typefaces aren’t copyrighted in the U.S. The only thing that is protected is the software: TTFs, OTFs, etc. [1] That’s why so many clones of popular fonts (and old metal type) exist.
These days the value in a font isn’t in the letterforms, it’s in the kerning, ligatures, variability, etc. which all flows from the software. It’s also where a significant amount of the labor in creating a typeface comes from. And it’s the thing that sets apart professional-quality fonts from many (but not all!) free ones.
If AI can write new font software by cloning bitmaps of letterforms _and_ get the kerning, ligatures, variability, etc. right… it’ll change the type foundry industry in a big way.
[1]: https://www.copyright.gov/circs/circ33.pdf
A recent doctoral thesis which looks at this:
https://lttrface.com/doctoral-thesis
Video from a talk from the ATD3 conference in Nancy which briefly explains the thesis https://vimeo.com/1059759506
I'm actively working on this with Fontweaver.com.
I do absolutely think integrating into the deeper features and functionality of fonts and opentype is what will make it work! Making it work well with existing font design software is also super important for typedesigners.
What are some high-quality free fonts that get kerning etc. right?
I see a project in there to generate kerning, hinting, etc, by rending text with commercial fonts then building a model against the rendered text.