How do I progress once I have a perfectly symmetrical /o/?

oribendor's picture

So I designed an /o/.

It looks fine to me, but it's perfectly symmetrical (with respect to a 180° rotation), and since I know the upper half tend to look bigger than the lower one, I thought perhaps I should apply some optical corrections.

So I examined some /o/'s of well known fonts, comparing them to their 180° rotations, and found out that indeed none was actually symmetrical.

So what's the procedure? Create a perfectly symmetrical /o/, then just move up the right- and left-most nodes a little? That sounds naïve, doesn't it?

How do you progress once you have a perfectly stmmetrical /o/?

---

My /o/:

Well known /o/'s (in red) with their 180° rotations:

cerulean's picture

In the examples shown, the differences are insignificant. I'd say your o is good to go. Visual center is more important to things that have a center, like s, x, H. So, for instance, if the o had a dot in the middle, you would have to raise it.

sardiez's picture

Hello,

Perhaps this is gonna sound more naive: my humble opinion is that your /o/ optically looks perfect. Maybe if you put the /o/ in context with other letters this issue could be more visible. Your typeface will not be better if it is done following the well known fonts. Karen Cheng says that there is not one correct way of designing typefaces, I cannot agree more.

Bye.

oribendor's picture

Well, I've examined more professional fonts, and none had a perfectly symmetrical /o/.

Why is it so, then? I mean, at least since fonts are being designed using computers, it's easier and more natural to create symmetrical /o/'s. So if they're not, there must be a reason, wouldn't you say?

Perhaps most type designers don't even start with symmetrical /o/'s? But how do they start, then?

I'd be grateful if other type designers could reflect on their experience with designing /o/'s.

nina's picture

One general thing I learned at lettering class back in design school is that the top-right segment of curves such as in the "D", but actually also the "O"/"o", may need to be a bit more pronounced, otherwise they can look deflated or even slightly backslanted. Not sure why this is. But like with all optical corrections, trust your eye; and it might be easier to stop staring at that "o", move on to other glyphs, view the "o" in context and if it rubs you the wrong way, then correct it. Not so much because others do it, but because you can see why.

One thing I've been wondering about for a while: To my eye, in symmetrical "o"s the top curve (near the x-height) tends to look very slightly thinner than the horizontal bit at the bottom, near the baseline. I have no idea why this is, and it depends a lot on the design at hand, but in your "o", I personally would move the topmost node of the inner contour down by a unit or maybe even two. I'd be curious if others see this too. (There are some fonts that do this correction, like Helvetica (1 unit in the Regular), or National (2 units).)
I'm not an expert at this, BTW, so take with salt.

Bendy's picture

Interesting question. I agree with Nina about the top right curve, and I usually draw it with emphasis there, and slightly raise the right hand extrema too. I suspect my /o/s are very unsymmetrical as I draw them by eye without any kind of flipping or rotating. I start by drawing two ellipses then push and pull the nodes and BCPs manually without measuring until it clicks.

Nina, I wonder about your second point. I wonder if it's something to do with reverse optical centring. To optically centre black parts of a glyph we raise them slightly above the geometric centre; here we might find the opposite works because we're dealing with a counterspace? I'll check out the fonts I've designed later to see if I've corrected the o this way.

Michael Jarboe's picture

One reason a letter might not be symmetrical is just by chance… at least with FontLab Studio where all nodes have to sit on whole coordinates what I've found is you can draw a circle using the shape tool and it can randomly draw out to be an even height/width when it needs to be at odd dimensions… so that the center nodes rest on the odd coordinate and either side is even… the more you scale said shapes up and down the more things can potentially become distorted… depending on the scaling and UPM size. You can see what I mean by drawing a few circles/shapes with 1000 or lower UPM and scaling it at extremes then doing a horizontal/vertical axis mirroring to see the symmetry get thrown off.

oribendor's picture

Very interesting comments, everyone.

nina,

the top-right segment of curves such as in the "D", but actually also the "O"/"o", may need to be a bit more pronounced, otherwise they can look deflated or even slightly backslanted

Interesting. Maybe it has something to do with us accustomed to calligraphy?

I personally would move the topmost node of the inner contour down by a unit or maybe even two

Yeah, I think I see it too. And it's not only the top curve, the right one too looks slightly thinner next to its left counterpart. Do you see it?

However, if you look at the three "well known" /o/'s I examined, you'll notice that the middle one has a thinner bottom curve, while the right one has a thinner top curve. I wonder if that means anything.

Bendy,

To optically centre black parts of a glyph we raise them slightly above the geometric centre; here we might find the opposite works because we're dealing with a counterspace?

Never crossed my mind, I think you've got a point!

Mike Jarboe,

I don't think you can actually attribute unsymmetries to rounding errors; they're too sharp for that.

Michael Jarboe's picture

Ori… yes your examples are much more extreme, just wanted to share my experience with the symmetry thing… it's interesting what you've found in your research, you'd be surprised what you find when you examine more well known typefaces in different ways…

Regards…

oribendor's picture

Yes, I think I will indeed examine more famous typefaces in different ways. It's quite fascinating, though has the potential of confusing you more. Beware of further questions! ;-)

Syndicate content Syndicate content