Learning a Manifold of Fonts (2014) | Hacker News Reader

Learning a Manifold of Fonts (2014)

vecg.cs.ucl.ac.uk

82 pointssebg16 comments

Heads up for anyone on limited bandwidth: The page loads over 25Mb of javascript in order to display anything.

Wow. Indeed. 44.2 MB for the whole page!

The morphing-letters demo is kinda cool though :-)

Yeah. Looks like their two dimensions correspond, left-right to serif vs sans-serif, and up-down some sort of bold/italic dimension?

This is good job. Having the ability to customize a font under the control of keeping it in the valid readable manifold.

Has Adobe discovered this? I be happy having this feature in Photoshop.

You would think that a web page about fonts would at least make a halfhearted attempt to use fonts that are somewhat readable.

I know we're not supposed to nitpick about formatting details and instead comment on the message itself. But when a page is about fonts, I think its choice of fonts is fair game. :-)

I tried to read the page in Chrome and Firefox on my Galaxy Note 4, and on desktop Chrome and Edge. It's barely readable in any of them.

Combine the spindly thin body font with the layout that doesn't reflow on mobile when you try to zoom in - what a mess.

The article would be so much more readable if it just didn't use a custom font at all and simply let the browser use its default font.

I was coming here to say this! It makes my eyes feel weird, it's dreadful

I concur.

It's not an article about fonts. It's an interactive experiment on typography.

I'm not sure that changes his argument much.

Hopefully I can be forgiven for thinking the article had something to do with fonts, considering that the word appears in the title as well as 22 times in the body. :-)

And I hope to be forgiven too for my poor wording, since I meant font readability :/

Not sure why that is the case, the font is a professional font by Adobe (also here https://www.google.com/fonts/specimen/Source+Sans+Pro). Maybe they are using a desktop font (without screen optimization) instead of an actual web font. You can see on the google fonts page that the font performs fine.

Having said that, their example for generated fonts are kind of painful to look at.

They use the 'lighter' value for font weight in the CSS, which is just absurd for body text.

This is the only thing I noticed before quickly closing the page. Completely unreadable.

The jargon in the type world is a tad different from what's being used here, it bothers me immensely. E.g. a font (file, lead type, etc.) and a typeface (design) are two different things. Type design and typography aren't synonymous either.

I know it's an experiment but I doubt it's likely that the EULA states that the fonts can be manipulated, e.g. you need written consent of Grilli Type if you wish to do this. I'm unaware of the policy in other foundries.

Neat experiment but I don't really see a use for this.

Reply on news.ycombinator.com