Font height question

Typogruffer's picture

Hey a fontlab newbie question here.
I have opened two two fonts names frutiger and Gill sans in fontlab to see their construction(cap height and ascender height.
But i noticed something different. while the cap height of frutiger is around 1400, but the cap height of gill sans is only 680. What is the ideal sizes(from ascender to descender)? i heard it should not be more than 1000? and also what happens if it is more than 1000. I didn't see any problem with frutiger.
Thanks
Typogruffer

John Hudson's picture

These fonts have different em unit definitions. My guess is that the Frutiger you are looking at is a TrueType version with a 2048 unit em, while the Gill Sans is a PostScript version with a 1000 unit em. The measurements of e.g. cap height are relative to the em, so in fact the Frutiger and Gill cap heights are almost identical.

hrant's picture

And/but looking for an "ideal" is misleading. Although in the past (as John's comparison helps reveal) there was more of a "pseudo-standard"* in terms of vertical proportions, these days we see more freedom there. Any "ideals" are more like general guidelines (which don't even always apply): make the caps shorter than the ascenders; make the descenders shorter than the ascenders; let the cap accents float a bit above the top limit.

* And within a given foundry (like ATF) often a strict standard.

As for the 1000 issue: it's a legacy limitation... which does not however make it totally irrelevant.

hhp

Typogruffer's picture

@John and Hrant: Thank you
Now just to make things more lucid, suppose i define "unit em" like 1500(i.e. from descender height to ascender height) and try to make a font. what difference does it make. What i mean will my 12 pt size be similar to the Frutiger 12 pt and Gill Sans 12 pt.
How is this size(1000 or 2048 or anything) actually define the length of your font(at various point sizes). How am i getting approximately the same length(at 12 pt) for both these fonts even though their height is so different while making.

hrant's picture

No matter how many units your Em is it gets scaled to fit in the "real" size. So 12 pt type will always fit in a 12/72 inch space... BUT "fit" does not mean what you might expect, since parts of a font (like the cap accents) can actually go over the upper or lower bounds (and it's the bounds* that are made to fit within the point size); also, you tend to leave some talus (AKA "internal leading") above the ascenders and below the descenders too.

* Which BTW is not the same thing as ascender + descender... :-)

hhp

John Hudson's picture

The em (also called the body height) is what is scaled, so it represents the height that will equal 12pt when the font is set at 12pt

Units per em defines the grid on which the outline is defined. The PostScript standard prior to OpenType was 1000 units per em. Most TrueType fonts were traditionally 2048 units per em. but this was not a requirement. In OpenType, either flavour of font may have any positive integer number of units per em, but for legacy and compatibility reasons 1000 is still recommended for PostScript OTF.

The ascender-to-descender height is often close to the em height, but that is not a requirement either: some fonts are 'cast small on the body', and a few may be taller than the em.

hrant's picture

a few may be taller than the em.

Like PMN Caecilia (if I'm not mistaken). Which I consider "cheating".

Also: if you want your font to look big (without cheating) you make the x-height big and/or you make the letters wide.

hhp

Typogruffer's picture

Thank you and your reply is very clear. I never knew about this scaling factor. But at same point sizes, some fonts appear Bigger and some smaller. Is it attributed to the difference in X heights? or some cheating(which i have know clue about), Hrant was talking about

Karl Stange's picture

In addition to the information provided by John and Hrant, you may find Karsten Luecke's detailed guide to font metrics a very insightful overview of this subject.

Like PMN Caecilia (if I'm not mistaken).

The version of this font I have access to (from the Adobe collection) seems to have fairly consistent metrics within a 1000 units per em setup. Unless I am missing what part is cheating.

hrant's picture

The apparent size of a (Latin) font comes almost entirely from the x-height region.

Caecilia: I guess I remember wrong; I thought its ascenders and descenders overshot the Em bounds (which is what I'm calling "cheating"). But here's an example I'm sure of: at some point sizes the bitmap renderings of the old MS Core Fonts exceed their nominal point size* (ascender to descender).

* I mean factoring in the 96/72 business.

hhp

John Hudson's picture

Differences in size between fonts at the same nominal size may be either apparent or actual. I would consider actual size differences to be differences between the ascender height and descender depth relative to the em. So, for instance, Monotype Perpetua is a type that is cast small on the body, i.e. the ascender-descender height is notably shorter than the em, and hence shorter than most other fonts. Apparent size difference is most likely, in Latin type, to be a result of difference in x-height and letter width, independent of the ascender-descender height relative to the em.

Nick Shinn's picture

Relative to the em, there is more variety in x-height than in cap height.
For obvious reasons.

So, because in U&lc text x-height is more indicative of apparent size than cap height, one does see clear differences in apparent size between different typefaces—at the same point size—even though, were their x-height to be equalized (by varying point size) and text compared, they might look the same size.

Syndicate content Syndicate content