I calibrated my MS Word settings so that I could print directly onto a sheet of notebook paper, saving me from having to hand-write a bunch of drafts in English class.
This app needs an upload functionality, where I can upload scans of my handwriting.
Where does the machine learning fit in? First guess is connection between letters but ill have to wait till I'm not at a keyboard to test this out.
They used to be myscriptfont.com and didn't have those premium features though. Don't know if it is better or worse.
But in general, I find it a bit hard to neatly write on the iPad. It doesn't matter much, I just use it to keep track of notes while programming and designing software. And I can read it myself just fine.
I think it's because the Pencil (version 1) feels more slippery than pen/paper while writing.
The advantage of writing on the iPad though: no more paper.
I honestly find it shocking that anything in the modern world still considers a signature as any form of authentication whatsoever. (Nobody actually compares the signatures, and yet somehow people see signing as meaningful.)
They're also useful for acknowledgements in paper work.
They aren't for authenticating that you're you, they're just extra information.
In our business, we very much do verify signatures. The primary goal is to establish whether the signatures are intended to be the same - not as some kind of forgery detector.
It looks like a combination between Elvish handwriting and the stuff you can find in the Voynich manuscript, but with all the elegance of Gregg Shorthand. I've somehow developed all kinds of weird habits. For example, beginning all my lines from the top and drawing down, or looping letter tails (like g and y) well beyond the line below them.
My ~90-100 WPM typing speed makes me feel okay about all this though because it's pretty rare that handwriting has much practical application these days.
It turns out that it looks reasonably nice in a font (except for my horrible keming) because most of the ugliness came from the characters all being irregular.
Curious to see if this one works better!
I specifically wrote double-decker As and Gs yet the the generated data has single story ones.
I don't even have a pen now on my desk.