# Mathematically calculating visual balance

Is it possible to mathematically calculate visual balance?

Below is an example of what I mean. Both images contain the letters "M" and "C," but are aligned differently. In the first image, the letters are centered and formatted to have normal kerning. In the second, the "C" is shifted to the right to improve the visual balance of the image as a whole.

Since the visual weight in the letter "C" is on the left and not centered as it is in the letter "M," the visual balance of the first example is off.

Would it be effective to mathematically determine a weighted average of the letters (white parts) and adjust spacings accordingly? This could be done by defining a function to represent the total height of the white bits at any given x value, then determining the weighted average (integral of x*f(x)) to find the visual "center" of the text and adjusting the text's position so that that "center" is in the image's center.

What your essentially asking is a special case of kerning. Systems that do automatic kerning have been done. This is called optical kerning in inDesign, you can find a few other companies have made their own tools. While its true that manual kerning make for better results in many cases than optical kerning there are situations where anything better than nothing is acceptable.

However, the algorithms behind different kerning solutions are usually trade secrets. But the lore knows a few patents and papers on how to automate kerning (see google scholar search for example). Overall this is a very complicated thing as the human visual system is incredibly complicated and kerning needs to account for very many things to be perfect.

So while doing a automated system may take quite a lot of development time to give even acceptable solutions, there is no doubt that such systems could not be made. In your case a custom system that may or may not pay itself back. It may be worth the investment in some special cases.

In any case if there is a formula its probably is hideously complex, and we dont know it as of yet. But that does not mean it does not exist. So currently you may be better off manually doing this in most cases that matter in any way.

• Notably, the context of this example of two separate (?) letters is very different from plain running text, which in turn is again different from all caps headlines. To top it off, actual size also matters. A 200 pt line on a billboard is kerned much tighter than the same text on an A5 flyer. – usr2564301 Oct 20 '16 at 17:55

typography itself has its rules to balance the types, ascendents, descendents, width, height, center, etc. balancing text is kinda about grasping the right spot to use as your reference frame for balancing. As in: if you're gonna use the center, the border, a special midpoint of the type, etc.

Due to the nature of the origins of StackOverflow, this sort of desire to reduce design to algorithms based on quack-like research analysis appears from time to time.

TRENDS exist in ALL THINGS SUBJECTIVE.

From "chaos theory" onwards, scientism has posited it's only a matter of time until all evolution, ebbs, flows and events can be divined. Until that fairy tale comes true, put some faith in the ideas, innovations, intuitions and instincts of designers.

Placement, spacing, weighting, colouring, contrast, disparity, inference and all the aspects of all things that make up `balance` vary...

They vary based on (amongst other things) platform, target, medium, purpose, audiences, cultures, times, interests, genres, niches, messages and contexts.

Trends, by their very nature, are liquid, and evolve and revolve from (again, amongst other things) influence, inspiration, invention and innovation.

Add to this the library of experiences in the utilisation and iteration on ideas from both originality and the twins of intuition and instinct (that lay at the very heart of a great designer) and you'll begin seeing the challenge for what it is. Impossible. Insurmountable.

But please don't forget the significance and impact of juxtaposition, revolution, deliberate disruption and nuanced visual harking and symbology deliberately designed to stir memories and references.

I hope you begin to understand design is a CREATIVE ART.

So, no. It is not possible to formulate algorithms that achieve "balance".

It is highly unlikely that any sense of visual balance (for the purposes of influence) is actually relative to any mathematical models of balance. And any moment in tides and trends when there is a correlation is pure fluke.

Attempting to mimic one tiny, subset of the trends that exist in design has not been achieved, (except for things like musicians modelling themselves on others) let alone anything that considers the more profound effects of mental cognition, recognition and ignition that created imagery is specifically designed to evoke.

And there's the keys to the kingdom of design. Graphic design has purpose - to influence humans. Code can't and won't ever understand the contexts and experiences of humans and how they shape their relationships, realities and reactions to that which they see and perceive.

If all the above fails to stir you, then remember this:

Letters, themselves, are not geometric and algorithmic in nature. They're evolved, imperfect, representations that are both comprised of and compromised in their formulation, form, function and functionality.

They're also not at all universal.

Hire a designer.

ps: Oddly, I think programmers have a vastly greater storage capacity for visual relationships, but that they perhaps archive in a non visual manner. Just look at the enormity of programming languages, frameworks, APIs and libraries they hold within to use and abuse, then the intertwining nature of code in even a simple app. Stunning stuff! That's a myriad of relationships and interlinking that's simply bigger than the biggest ever infographic. And they hold this in their heads!!!

• While I agree in most part with your answer, I think the tone is sometimes a bit harsh and condescending. Also, there is no mention of code or programmers in the question, so your conclusion is a bit 'off the bat'. – PieBie Oct 20 '16 at 17:56