Optical Vs. Metric Kerning

Report · Sep 22, 2008

What are the differences are between "Optical" and "Metric" kerning in InDesign? In what cases would you use one more than the other - is there a hard rule to when you would select either? Or is it just personal preference? Any help is appreciated.

Thanks,

D

Report · Dec 21, 2008

>I think it is generally as least as good as metrics, often superior and definitely the way to go unless you want to slave over every kerning pair to get it just so.

But the point with metrics is that a font designer has already slaved over at least a good portion of the kerns to get them "just so". And for those kerns that are automatically generated, if type desigenrs thought that ID's optical kerning was better than the kerning routines they previously used, they would use ID for kerning. To me, for fonts from reputable foundries, it seems obvious that either you get no improvement (if only auto kerning was used) or you lose the advantages of the designer manually checking the kerning. I'm in agreement with Richard here.

Report · Dec 21, 2008

My experience agrees with Dave's. I've never felt any resource "hogging" due to the use of Optical Kerning.

This argument seems to me a bit silly. It's all about aesthetics and even more about the kerning pairs in each and every specific font. Some fonts have good built in metrics, and others do not. It is all quite subjective, and there is no right or wrong answer....

Report · Dec 21, 2008

Improvements come from the fact that optical kerning keeps on working when glyphs from different fonts are next to each other, or different styles of the same font or even different sizes, and, of course, it works for every pair in the font, not just the ones the designer bothered with. Frankly, from what I've seen of designed kerning metrics, there's a lot of energy expended for little real value compared to optical kerning. InDesign's optical kerning didn't exist when a lot of these fonts were designed.

But to each his own.

Dave

Report · Dec 21, 2008

Fonts are designed (for the most part) to be used at body copy size and the metrics are set accordingly. It's when they are used at larger size that the metrics fail visually and one resorts to optical.

Got this from Graphic Design Forum:

<<Every font has in-built data on kerning pairs between its letters. For most cheap/free fonts, though those kerning pairs are somwhere between awful and non-existent. For a font from a major type foundry, like Trajan, there will be an exstensive set of kerning pairs. That's what you get if you apply metrics. Those kerning pairs were created by an expert font designer, and should get you good results...

at the size they were designed for.

Which is generally around body copy sizes. However, kerning for text at display sizes is quite different to that at body sizes, and the metrics at that point are probably way off. That's where the optical setting comes in. There, Indesign tries to guess the correct kerning based on optical weights of the letters. It will usually get you far closer to correct than metrics at display sizes, but under most situations hwere you'd use optical, you should probably expect to need to do at least some manual kerning as well.>>

What are the "body" sizes those of you applying optical kerning to paras? I don't consider anything beyond 11-12 point body copy.

Maybe it is subjective but I think optical kerning applied to body copy is a lot of unneccesary busy work. My gut feeling is anyone applying optical kerning to an entire paragraph is using it in lieu of proper tracking which is the spacing between all letters. Again, this can be adjusted in ID prefs or within style sheets.

Report · Dec 21, 2008

>unneccesary busy work.

How is setting a paragraph style to use Optical rather than Metric kerning "busy work"? It is a trivial one-time setting...

Report · Dec 21, 2008

>Improvements come from the fact that optical kerning keeps on working when glyphs from different fonts are next to each other, or different styles of the same font or even different sizes

These are very specific circumstances and I don't see that they justify the blanket assertion that optical kerning is "often superior and definitely the way to go". Frankly, when I do get glyphs from different fonts adjacent to each other, I prefer to kern them manually.

>InDesign's optical kerning didn't exist when a lot of these fonts were designed.

No, but Kernus and other kerning programs did and I haven't heard that they've been dropped wholesale for ID's kerning.

I don't recall this sort of devotion to optical kerning when it was present in PM as expert kerning. I've said it before and I'll say it again, if it were called "computer kerning" and metrics "designer kerning", I think that would have an effect on people's preferences. I strongly suspect that most people who always use optical have not done anything more than cast a cursory eye over the results before and after and simply think "optical" sounds more designer-ish than "metrics".

Report · Dec 21, 2008

In PM's day, the computers were slow and Export Kerning had to be applied as a user initiated process; if memory serves, it didn't survive editing, so you had to do it as the last thing you did. So, it was nowhere near as convenient and optical kerning is now.

A few years ago, I did a lot of careful looking at optical vs. metrics and concluded that each had its strengths and weaknesses and the the weaknesses of optical were easier to overcome.

Dave

Report · Dec 21, 2008

>I did a lot of careful looking at optical vs. metrics and concluded that each had its strengths and weaknesses

You see, I'd be interested in reading those lists, rather than just assertions that optical is best. To me, the weakness of each is that they rely on algorithms to generate an aesthetic result (something not best served by programs), but metrics has the added avantage of a trained human eye being run over the pairs.

>However, kerning for text at display sizes is quite different to that at body sizes, and the metrics at that point are probably way off.

Richard, I don't follow this. Why does kerning have to change with point size? If there is a correct visual amount of space between letters, why does this suddenly become wrong when you increase the point size? Yes, there is the separate issue of tracking, but that is a text-wide setting.

Report · Dec 21, 2008

It was five years ago and I've long since lost the specifics. But it came down to: 1. Optical kerning, in effect, has built-in tracking, so when you use it for different sizes of text, it automatically adjusts without the need to tweak the tracking. 2. Optical kerning works on all pairs and only occasionally produces a bad pair. Metrics only apply to the pairs the designer attended to. 3. Metrics in some fonts (and I'll grant I might be a bit over-sensitive to Gill Sans, which is the worst offender) cause spaces to disappear at the end of sentences when the next sentence starts with a T or W. 4. There is no discernible speed difference between metrics and optical (I mention this only because Richard keeps implying that your computer is all but consumed by the processing power that optical kerning allegedly uses). 5. In the work I was doing at the time, it was frequently the case that I was working with text that had mixed styles, mixed fonts and mixed sizes, so the fact that optical kerning kept on trucking in those cases was relevant to me. 6. In the final jobs I did with PageMaker, I went so far as to generate a PalatinoPDS variant of Palatino to address some of its metrics weaknesses. I forget which changes I made now, although I do recall that <open-double-quote><capital A> was one that look plain awful in the font because it was missed by the designer. Dave

Report · Dec 22, 2008

In reply, I'd say that:

(1) I prefer to set tracking explicitly by style, not hope that some algorithm is going to come up with the right amount.

(2) I don't have the same faith or experience in near perfect optical kerning. In general, I find it sets type too loose.

(3) As I've said before, this isn't an error with the metrics - it's simply true optical kerning, where the space between letters is balanced. You may not like it, but that doesn't make it an error.

(4) I don't see any major speed loss with optical kerning, and this is not an argument I advance against it, though in scripting I accept it may be an issue.

(5) In such cases, I can see the value of optical kerning for those who can't or don't want to manually kern, but most discussion and advocation of optical kerning in these forums seems to centre around just turning it on for long stretches of standard text.

(6) I see such instances as an argument for a kerning editor, not optical kerning.

Report · Dec 22, 2008

re: (2) I don't have the same faith or experience in near perfect optical kerning. In general, I find it sets type too loose. But you can always overcome that with tracking <grin> (3) As I've said before, this isn't an error with the metrics - it's simply true optical kerning, where the space between letters is balanced. You may not like it, but that doesn't make it an error. Of course it's an error. It's a sucky error so bad that the person who made the mistake should have been fired forthwith. Fortunately, it is an error that is largely restricted to Gill Sans and so not a good general argument. I guess I must have been using Gill Sans on some job back in the 90s when this first came to my attention using PageMaker. Dave

Report · Dec 22, 2008

I think the main issue is people are using optical kerning when what they really want is optimal tracking.

Here's a link to Michael Murphy's website The InDesigner and two webcasts about type and tracking. Very talented, articulate guy.

Report · Dec 22, 2008

>Of course it's an error.

No, I believe it's an intentional design decision made to reflect a style of typesetting that used to be more common. To call it an error means I could call any of your work I aesthetically dislike an error, which I'm sure you would disagree with. It's also akin to saying the old typesetting habit of putting em spaces after full points is an error just because most people don't follow that style anymore. I haven't seen many books set these days with diminished space after full points but I have seen them.

>I think the main issue is people are using optical kerning when what they really want is optimal tracking.

I haven't seen any evidence that this is the case. If people just want to track text, they have far greater control with the tracking setting and I don't see how they'd miss that and instead chooose optical kerning.

Report · Dec 23, 2008

I agree with Dominic about Gill Sans. The font is spaced pretty much the way Gill designed it, and if you check out the avant garde typography of the period, you'll see plenty of similar spacings.

Yours
Vern

Report · Dec 23, 2008

In my brief google, I couldn't come up with a better example than this:

http://en.wikipedia.org/wiki/File:Sir_Harry_Johnston_memorial_plaque.JPG

it's Gill's stonecutting rather than typesetting but it does illutrate the point I (and Vern) are making. Note especially the last line, where spaces are omitted where commas are present, because the white space around the commas was thought sufficient to break the text.

Report · Dec 23, 2008

So why then is this only applicable to sentences that start with T and W?

I have no objection at all to the negative 125 applied to period-space. It's when you combine that with the negative 100 applied to space-T that the problem rears.

It's worse in Gill Sans bold where the combination amounts to negative 250.

Dave

Report · Dec 23, 2008

>So why then is this only applicable to sentences that start with T and W?

It's not. But when a sentence starts with a T, the space under the bar of the T was considered sufficient differention (as I say, "true optical kerning"). Same with W and A and V and Y. Those letters have space to the left because of their design and are kerned more tightly than, say, a I or an H.

Report · Dec 23, 2008

And that, in my opinion, is the mistake made by the kerning -- although you are saying that it is not a mistake but deliberate -- taking two bites out of the same cherry so that space all but disappears.

Just because the type designer thinks this is a good idea is not going to persuade me that it is.

Dave

Report · Dec 23, 2008

Reread my posts - I'm not trying to persuade you that it's a good idea and I've never even said that I like that style myself. But it's not an error. That's like calling an ID feature that works as designed a "bug" simply because you would rather it worked a different way. And, I think that such differences of opinion between the designer's and the user's kerning preferences are best sorted out by a kerning editor, not an optical algorithm.

Report · Dec 23, 2008

It is worth mentioning that with CS4, this can be resolved using a GREP style if you want to stay with metrics for the rest of the font.

And, fwiw, there are a number of as designed features in InDesign for which I would apply the label design bug. In fact, it's always been my opinion that design bugs are worse than regular bugs because they actually intended that it work that way!

Dave

Report · Dec 23, 2008

Then they cannot, by definition, be bugs.

"A bug, also referred to as a software bug, is an error or flaw in a computer program that may prevent it from working correctly or produce an incorrect or unintended result."

Yours
Vern

Report · Dec 23, 2008

Of course they can. Designers are just as capable of making mistakes as are programmers. But they do deserve a different name. That's why I call them design bugs.

Dave

Report · Dec 24, 2008

If you're using optical kerning universally you're implying that every font is poorly designed which is a load of bs.

Report · Dec 24, 2008

>because they actually intended that it work that way!

Not only that, but there is at least one engineer interested in defending each of them :-)

Report · Dec 24, 2008

Richard,

No. My use of it is primarily driven by the fact that it works universally and not just selectively. That, to my mind, is the main problem with metrics. There are only so many metrics included in each font.

Dave