10

Ask DN : What are the core problems while working with Type for any project?

over 1 year ago from , National Institute of Design

What is the most annoying, time taking process while working with Type with your projects(branding, UI Design anything)?

When I say Type, it could be Type Selection, Detection, Manipulation or Pairing anything.

I'm trying to understand the core-problems; once that figure out I'll try to address those problems using Machine Learning to see what can be done to resolve those problems.

PS: I want to assist designers in their daily creative process by using machine learning as a technology; starting it with Type.

14 comments

  • Richard BruskowskiRichard Bruskowski, over 1 year ago

    Working with type where? In design tools, I'd say consistently applying the right typographic styles to the right texts for the right viewport width and so on is more of a hassle than it should be. This is valid, once the typographic system is set for a specific project. That is settled with a few of lines of code for developers but often a jungle of text style presets, tables and/or cerebral matter for designers in most of graphic design tools. I'd love to see some computer-aid there. E.g. a plugin, which guesses and applies the intended font-size, color, spacing, …. In the creative process of choosing, selecting type, pairings? That's way less painful and I would not want to rely on algorithms there as much as I'd expect them to be potentially limiting in some ways.

    3 points
    • Neel KadiaNeel Kadia, over 1 year ago

      Thanks Richard for the comment. So basically are you talking about applying specific typography styles automatically by looking at the content on the artboards can be interesting way to look at? I'll look into that and take this in account while going further.

      and I want to assist designers in their creative process also. Can there be any other area, particular in type, where you think machine assistance can do your work a lot more easy?

      1 point
      • Richard BruskowskiRichard Bruskowski, over 1 year ago

        Not so much by looking at the content but by looking at the hints I as a designer gave. So if I placed a text layer with 40px font size, #000000 black in Roboto on an artboard that belongs to project X for brand Y, my friendly typographic assistant could suggest to change it to 36px, #1a1a1a, Product Sans, -.01em character spacing and 1.1x line-height because that would be the closest styling that's correct according to the typographic system of the project (and brand). Basically linting for designers. Actually not sure if ML is needed for that, but maybe ML could help to guess the intention of the designer more accurately.

        In the creative process, helping to find close alternatives could be helpful. E.g. if I choose a font that is not available or too pricey for a project, quickly get some high quality alternate options with means to compare pricewise. Or finding a font with squishy search terms, like "soft", "warm", "friendly, but not too friendly".

        1 point
    • Roland IllésRoland Illés, over 1 year ago

      This launched not long ago, and seems to be what you're looking for: https://stackswell.io/

      3 points
  • Phil RauPhil Rau, over 1 year ago

    Well, this is probably only relevant for a really niche area of design, but I could imagine that machine learning could be used to help type designers (those who create their own typefaces) craft both the vector letterforms and help them tune the kerning of the font.

    Perhaps a typeface designer could give the computer 5-10 letters, and the computer could use machine learning to generate the rest of the typeface?


    I'm not sure if typography is the best place to start here. I wonder if a computer might be better at cropping and enhancing images, or helping with web layouts.

    3 points
    • Neel Kadia, over 1 year ago

      Thanks, Phil! I second you for the idea; but does it limit the creativity of the Designer? I've asked few foundries about the same, their response was like, for typographers typeface creation is like an Art; having a tool where you drop few letterforms and computer create rest of the missing letterforms won't be their first choice.

      0 points
      • Phil RauPhil Rau, over 1 year ago

        For sure. Well I mean, I feel like many designers would feel the same way about machine learning replacing any aspect of their job. I'm just not sure if machine learning is able to replace a tool, I think instead its better suited to (eventually) replacing some things the designers do.

        0 points
    • Richard BruskowskiRichard Bruskowski, over 1 year ago

      Perhaps a typeface designer could give the computer 5-10 letters, and the computer could use machine learning to generate the rest of the typeface?

      I like the idea. Not to actually produce fonts, but eventually to enhance the ideation and decision making phase in some respects by letting the type designer quickly have a preview of the potential of his ideas.

      0 points
  • Account deleted over 1 year ago

    One thing that annoys the hell out of me is when you get a design looking all nice in sketch, then code it up and all the weights look about 1 weight heavier than they actually are, so you have to go through your code and change everything again.

    1 point
    • Dylan SmithDylan Smith, over 1 year ago

      That's an antialiasing problem, not a font-weight problem. Look up how to do antialiasing in CSS. One bit of code should sort it out project-wide.

      1 point
      • Account deleted over 1 year ago

        Yeah I've tried different AA settings, it always seems to be a little bit fatter/heavier in browser compared to sketch no matter what setting I put it on. It's irritating.

        0 points
  • Joe AlfonsoJoe Alfonso, over 1 year ago

    I recently made the switch to FIgma for some work due to a more robust typeface and Opentype toolset. Having the options for vertical alignment, proportional and tabular characters is really nice too.

    1 point
  • Theresa MershonTheresa Mershon, over 1 year ago
    1. Licensing for web - inconsistencies in how foundries present licensing options and some restrictions on use that make it difficult (e.g. a license that does not allow you to legally use the hosted font in a form field because it can be edited by site visitors is one of the really stupid limitations I've seen in some EULAs)
    2. Website performance. Fonts have to load pretty high in the waterfall and therefore always get flagged in performance audits. While I am grateful to be able to use web fonts at all, there's a lot that could be improved in the way we serve web fonts.
    3. Inconsistent baselines across fonts and browsers. Having to add css to visually fix vertical centering.
    0 points