Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Simon Cozens

About

Username
Simon Cozens
Joined
Visits
2,926
Last Active
Roles
Member
Points
355
Posts
264
  • Re: Units per em

    I did talk with ORA about this. My contact in O’Reilly asked around to see if there would be any chance of this, but “so far the answers have all been ‘no’.” (Not sure why - that’s all I got.)

    On the other hand I don’t think much of the material has aged well. Chapters about TeX fonts and Mac OS 9 aren’t particularly relevant any more, and a book like that today should have much more coverage of global scripts. That’s why I decided to start again from scratch, and that also gives you the advantage of a consistent writing style.
  • Re: Units per em

    I've spoken to the folks at O'Reilly and they are not planning on doing a new edition of Fonts and Encodings. The cost involved in getting the book through production (obviously a book about fonts required more specialized production than their usual process) was not seen as worth the benefit given fairly meagre sales.

    I'm very slowly working on a book called Fonts and Layout For Global Scripts, which I hope to be a more modern (and less Latin-focused) replacement, but recently have got sidetracked writing a different book for my day job. Contributions to Fonts and Layout are welcome.
  • Re: Plex; IBM's new font identity model

    This statement is empirically testable. What’s the evidence base?
    Is it? It's not as if we can actually do an experiment, setting up two parallel timelines with and without free fonts.
    No, but we can see whether or not the growth of free fonts (proxied by e.g. number of fonts on dafont over time) has led to a corresponding decrease in sales of commercial fonts (proxied by e.g. total revenue on Myfonts or similar over time). If you're trying to assert that free fonts pose a danger to paid fonts, that's the sort of evidence you're going to need to show.

    Otherwise it's all just anecdotes and supposition.
    Ben Blom said:
    I do not have any empirical data, whether the font market is actually in decline, or not. And even if I would have such data, it wouldn’t be very clear. It’s like the climate. If one year is unusually hot, it’s not clear whether that is a manifestation of climate change, or an exceptional year. As with climate change, there may be a “tipping point” in the font market, when it will be clear for all, that something has fundamentally changed. Until then, many people may continue to suggest it’s business as usual in the font market.
    This is not how climate science works. It's precisely because we can plot historical data that we can determine real trends rather than outliers. So having historical data about sales certainly would help your claim about free fonts destroying the market. (Or, I suspect, hinder it.)
  • Re: Exporting fea file

    Fonts And Layout gives three ways to do this: https://simoncozens.github.io/fonts-and-layout/features.html
  • Re: I trained a neural network to kern a font (mostly)

    I came back to this yesterday and threw another couple of hundred fonts to learn from and gave it lots of time on some very expensive hardware. It claims to be up to 92% validation accuracy, which certainly sounds good. I've uploaded the latest kerning model and added some instructions: https://github.com/simoncozens/atokern/blob/master/README.md

    I would be interested to hear feedback on whether it's unusably off the mark in its predictions; off-the-mark but promising; reasonably close to accurate; or whatever.