22 comments

  • Marc EdwardsMarc Edwards, 4 years ago (edited 4 years ago )

    Take a look at the image below of a square next to a circle. When the circle is mathematically the same height as the square it appears to be too small.

    Because they are not the same area. If you want them to be the same visual weight, there is always a mathematical solution, you just have to find it.

    Circle diameter = sqrt(area/pi) × 2

    For an area of 1, that’s 1.128379.

    If you want your circles to look the same weight as your squares, scale them by 112.8379%.

    Most people think all you need to do is change the colour to white and you’re done. But… that’s not the case. Something weird happens… the logo actually looks ‘fatter’. The contrasting difference has made the exact same logo look like it’s put on some weight!… weird right? This illusion is called Irradiation Phenomenon.

    I'm not disputing Irradiation Phenomenon and the other tricks our brains play on us, but it is also worth paying some attention to ink bleed in printing, and shoddy antialiasing for screen design. There’s sometimes a technical reason why things look wrong, and I’d tend to try to observe those before making guesstimate adjustments to your artwork.

    13 points
    • Jacob TaylorJacob Taylor, 4 years ago

      What designer, realistically, is going to spend their time working out a formula like that?

      3 points
      • Dirk HCM van BoxtelDirk HCM van Boxtel, 4 years ago

        Well I don't want to say "the ones that give a fuck about their work", but; the ones that give a fuck about their work.

        When I read stuff like that, my brain picks it up because it's information that will help me become a better designer. From here on out, I'll remember 112%. Or 110% at least. Which makes me a better designer. Even if it's just by 0.1%.

        C'mon man. Use your skills of empathy. A designer's strongest asset. Try and think of a reason why you SHOULD use a formula like that. A single case where it would be useful.

        I'll do one. Maybe when you're working on a logo for a big important client that will lead you to more work. And you just can't get it to feel right. That 112% might be the saviour. And then during your presentation to the client, when they raise a point about the circle and the square not being the same size, you can tell them about visual weight and even give the percentage! They'll be crazy impressed and hire you for that oh-so-important phase 2 where you get to do the website! BAM. Difference maker.

        Don't ever discount theory.

        9 points
      • Marc EdwardsMarc Edwards, 4 years ago

        What designer, realistically, is going to spend their time working out a formula like that?

        I didn't work out the formula, it’s well known and only took a few seconds to find. The difficult part would be knowing that you want to balance by area, rather than by lining up using guides.

        I think the question is more: If you see an issue with something not balancing, you should question why, rather than just saying “I’ll fudge this a little until it’s kind of correct”.

        I hate kind-of-correct. Kind-of-correct is not correct.

        8 points
        • Jacob TaylorJacob Taylor, 4 years ago

          I hate kind-of-correct. Kind-of-correct is not correct.

          Depends what your definition of 'correct' is. If it looks visually correct, I couldn't care less if it was mathematically correct.

          If I care about maths I'd be a developer.

          Wasn't that the whole point of the article?

          0 points
          • Marc EdwardsMarc Edwards, 4 years ago

            I’m surprised at your comment. I don’t know how I’d design without maths.

            And, ignoring the maths involved doesn’t mean it don’t exist.

            2 points
    • Dominic SebastianDominic Sebastian, 4 years ago

      I know you're written pretty extensively on this area Marc, but have you found any other good sources of truth for these kinds of design hacks? Or is your process always - that doesn't look quite right... search for why. Rinse and repeat ?

      0 points
      • Marc EdwardsMarc Edwards, 4 years ago

        I haven’t seen a good source of help on this, but it would be great if someone documented common design issues (like overshooting curves) with some research and maths for possible solutions that can be replicated in our work. Maybe someone has already done that?

        It may not have come across this way, but I’m all for adjusting things as you see fit, or hunting for maths involved. Both are good and I definitely wouldn’t expect everyone to turn a 5 min design task into a thesis and search for some ultimate answer to whatever they’re trying to solve.

        But, I also don’t like hand-wavy “optical corrections”. I don’t think suggestions like that help beginners or experienced designers. There’s nothing concrete you can replicate in your own work. “Oh, I just move things around until they look good” isn’t helpful advice.

        0 points
    • phil helmphil helm, 4 years ago

      Actually the author is discussing the visual height (not weight) of the circle compared to the square.

      When both are the same height, the circle appears shorter than the square and is increased in size to give it the appearance of being the same height as the square (where in fact it is taller).

      The weight of the circle is still less than the square, using your formula the circle would appear much taller than the square.

      The author is explaining that the visual height / weight does not necessarily follow the mathematics and adjustments by eye are often needed.

      1 point
      • Vlad Danilov, 4 years ago

        It’s totally possible to solve this mathematically, e.g. make overshoot areas equal to width✕1 pixels. The problem is there’s no demand for such tools. So in reality it's easier to do it by eye.

        1 point
        • Marc EdwardsMarc Edwards, 4 years ago

          I’d love a tool that had some knowledge of these things. Even just some smarter alignment and distribution tools would help (ones that don't blindly follow axis aligned bounding boxes, but are a little smarter).

          1 point
      • Marc EdwardsMarc Edwards, 4 years ago

        Sorry if I misrepresented the specific case in the article. You’re right.

        I definitely think it’s solvable with maths or research or both. “Adjustments by eye” are so imprecise. At the very least, it can be good to incorporate a curve overshoot guide in a grid system, so all your curve overshoots match.

        1 point
  • phil helmphil helm, 4 years ago

    A complimentary article on design adjustment based on optical perception. https://medium.com/@lukejonesme/optical-adjustment-b55492a1165c

    1 point
  • Thom StoodleyThom Stoodley, 4 years ago

    Originally published at logogeek.uk on September 4, 2016.

    Funny that

    0 points
  • Jacob TaylorJacob Taylor, 4 years ago (edited 4 years ago )

    Design isn't math. Design doesn't have a 'correct' solution. When you are working in the startup industry, where everyone likes things to be quanitified, this can be a harder truth to deal with than in traditional design jobs. Sometimes things are a bit more vague or fuzzy and we need to acknowledge that.

    0 points
    • Dirk HCM van BoxtelDirk HCM van Boxtel, 4 years ago

      Design IS math. Design DOES have a correct solution. Even if we can't (yet) comprehend it.

      How do you explain the leaps and bounds we've come since the start of the web? It's because we've discovered stuff. Because there ARE truths out there that will help us become better designers.

      Design is not art.

      ps.: I just realized you're the same person I typed the other post to. This is not personal. It's just that I think you're missing a very important point: we CAN measure almost everything, and this is how we move forward in the world.

      4 points
      • Marc EdwardsMarc Edwards, 4 years ago

        we CAN measure almost everything, and this is how we move forward in the world.

        Yep. And, if you think 112% looks wrong, there’s likely a reason for it, and yet more maths that can help.

        2 points
      • Jacob TaylorJacob Taylor, 4 years ago (edited 4 years ago )

        How do you explain the leaps and bounds we've come since the start of the web?

        Developments in technology, and having actual designers do the work, rather than devs.

        The basic rules of graphic design have existed since the Bauhaus. Just because it took a while for them to be applied to digital design does not mean they are a recent development.

        Design DOES have a correct solution

        A correct solution? I don't think so. If that were the case, every designer would be producing identical work. Yes, there are definitely rules that define right and wrong in design. But there's no such thing as a measurably correct solution. Design is subjective.

        0 points
        • Dirk HCM van BoxtelDirk HCM van Boxtel, 4 years ago

          This is what we're saying: design IS measurable.

          Otherwise, explain why A/B testing works. Explain why eye-tracking works. Explain why research gives us new results.

          Saying we're stuck in Bauhaus is like saying "design was invented and never changed since the Egyptians". Come on man, there's been tons of progress, we've learned SO much!

          Design is about getting a message across. Getting a message across is an I/O operation. You either do it or you don't. Which means it's measurable. Which means there's correct, and incorrect solutions.

          0 points
    • Marc EdwardsMarc Edwards, 4 years ago (edited 4 years ago )

      Things are vague and fuzzy, but not for the reasons you describe — there’s many paths you can choose, and those are decisions can be fuzzy.

      But, if you want to do something like weigh two elements visually, there’s tons of maths at your disposal that will do a far better job than fudging things until stuff looks right. This is even more important as the project scales. What if you didn't have 2 icons to match, but 200? 2000? You’d better bring some rules, grid systems and maths if you want to do that well.

      1 point
    • Duncan RussellDuncan Russell, 4 years ago (edited 4 years ago )

      Design isn't math

      I hope you never have to design anything inherently functional ;)

      2 points