Accents and Metrics

stimopo's picture

Hello, I am in the middle of designing my first font and have a little stumbling block in terms of the metrics measurements.

I have managed to insert the 'correct' values for most glyphs but I am unsure of what to do with accents such as circumflex, tilde, breve, cedilla, caron etc. I created these with their corresponding letterfoms initially and so Fontlab autofilled these spaces hence leaving the metrics values rather large. I am unsure of whether to leave them large as is the case with other fonts I have studied or manually assign some values to them? Hope I make sense?

Thanks for any help offered...

blank's picture

I center the marks on an en (500 units), as do many other designers. But I also see them centered on common widths that vary from font to font, and in some cases, with variable widths. So maybe some of the more experienced designers can explain a little more.

charles ellertson's picture

Far as I'm concerned, you're on the wrong track. You need to know the difference between SPACING MODIFIERS and COMBINING DIACRITICALS. These are different characters in Unicode, even if they are the same glyph. OpenType presupposes Unicode.

Just remember that as far as Unicode was concerned, having a codepoint for any accented character was an unfortunate circumstance, done for legacy purposes only. Some people are beginning to prepare texts as Unicode intended, even when a legacy character is available in most fonts -- such as aacute.

So your first job is to not muck up the proper spacing of the characters. Your second job is to make up whatever precomposed accented characters you want, without buggering up that proper spacing.

stimopo's picture


I am not sure I follow. I have attached a photo of my typeface to show you what I mean.

The character on the left (including an accent) has been spaced correctly. ie. the side bearing measurements are correct. The accent on the right (on its own) is the one in question. How do I work out the side bearing values to give all single accents? (Like cedilla, ogonek, circumflex, breve etc.)?

charles ellertson's picture

A combining diacritical

should have zero width. You can locate that accent anywhere (with a plus or minus sidebearing value), but the character needs to have zero width.

Spacing modifier characters

should have their natural width, so they appear right when used in a word -- an example, not from the list of "accents," is the "raised comma," as is sometimes used for a romanized hamza, which is properly U+02BD, not an "apostrophe," U+2019. The two different characters will likely need a slightly different width, and different kerning.

Same with different uses of "accent" characters, such as the "acute." As I said, you will find the same glyph in several places, but they are different characters, and need to be spaced for that character's proper use.

Or, maybe you are asking about "what's the best width so I don't have to do as much work?" But how much work you have to do to get the accent right on precomposed characters is your problem (i.e., the type designer's problem). Not a proper excuse for mucking up the character's widths.

JamesT's picture

For my typeface, I centered the combining diacritics based on the metrics for the /o/ or the /O/ depending on whether they were for caps or lower case. For the rest of the diacritics, I used the same size as I did for the tabular figures.

stimopo's picture

Charles_e – I am obviously trying to find out how to do this, I am not trying to avoid doing the work.

Thanks for everybody's comments so far.

Syndicate content Syndicate content