Glyph Transformer scripts and the Private Use Area

FontCreator - Font Editor FontCreator - Discussion
1

I notice that the Glyph Transformer has a number of supplied scripts which add glyphs into the Unicode Private Use Area.

Some of the glyphs added by the Ligature Collection.xml script are in the Private Use Area.

The Private Use Area mappings used in Ligature Collection.xml are a selection from those specified by the Medieval Unicode Font Initiative. The webspace of the Medieval Unicode Font Initiative is http://www.mufi.info and various documents are available there.

Yet the origin of the mapping choices for the small capitals is not stated.

I am thinking of trying to make a font with various swash glyphs for characters in the Private Use Area, both for use now in non-OpenType-aware applications and so that the artwork is stored in case OpenType-aware applications become more available and OpenType versions of my fonts become made. Some would be usable with regular roman fonts as well as with italic fonts, yet many would be more suitable for use with italic fonts only.

I realize that I could make my own private mapping allocations, yet I am hoping to use mappings which would be common with fonts from other designers, so I am wondering if something could be agreed and an xml file for the Glyph Transformer of FontCreator produced. It would be helpful to me if the range U+E700 to U+E7FF were not used in such allocations as I use that range for various ligature glyphs with my own code point allocations already.

I have been looking at the selection of swash glyphs, including those for endings and those for swash ligatures, in the following pdf document about the Arno font.

http://store1.adobe.com/type/browser/pdfs/ARNP/ArnoPro-Italic.pdf

And yes, I am wanting to include an ending version of the Esperanto ĥ character as mentioned in the thread entitled “An unusual glyph of an Esperanto character in the Arno font”, complete with a kerning pair with the closing single quote character, in one of my own fonts, so that I can try to produce some hardcopy prints.

http://forum.high-logic.com:9080/t/an-unusual-glyph-of-an-esperanto-character-in-the-arno-font/2072/1

Yet I am also studying some of the swash characters being used in some of the Festival Books mentioned in the following thread.

http://forum.high-logic.com:9080/t/british-library-treasures-in-full-webspace/2114/1

William Overington

18 June 2008

2

A helpful-yet-now-deleted post in this thread contained a reference to the following document.

http://www.high-logic.com/fontcreator/Using%20Glyph%20Transformations.pdf

From that document I became aware of the significance of the CompositeData.xml file in some Glyph Transformations.

Looking at the Private Use Area mappings used for various features in the http://www.high-logic.com/fontcreator/Using%20Glyph%20Transformations.pdf document and wishing to avoid clashing with them, I have now thought out the following list which I think do not clash.

Swash Capitals, usually for Italic fonts, though could be used for alternate capitals for Roman fonts.

U+E291 - U+E2AA 58001 - 58026 A - Z Swash Capitals
U+E2AB - U+E2B0 58027 - 58032 Esperanto Swash Capitals Ĉ Ĝ Ĥ Ĵ Ŝ Ŭ

U+E313 - E324 58131 - 58148 Lowercase alternate ending glyphs.

The characters listed are a selection of those included in the
http://store1.adobe.com/type/browser/pdfs/ARNP/ArnoPro-Regular.pdf document, together with a ĝ as well.

U+E313 58131 a
U+E314 58132 d
U+E315 58133 e
U+E316 58134 g
U+E317 58135 h
U+E318 58136 m
U+E319 58137 n
U+E31A 58138 r
U+E31B 58139 t upper
U+E31C 58140 t lower
U+E31D 58141 u
U+E31E 58142 z
U+E31F 58143 è
U+E320 58144 é
U+E321 58145 ê
U+E322 58146 ĝ
U+E323 58147 ĥ
U+E324 58148 ŭ

An implementation of the above lowercase alternate ending glyphs in a font which may be used to set Esperanto poetry could usefully include, for each of those characters except è, é and ê, a kerning pair of the character with a close single quote character, so that the alternate ending glyph followed by an apostrophe is stylishly set. That may perhaps not be necessary for a, e and u, yet I am unable to say that it is certainly not necessary, so it may well be a good idea to include the kerning pairs in the font.

U+E325 - E326 58149 - 58150 Lowercase alternate beginning glyphs

U+E325 58149 v
U+E326 58150 x

U+E327 - E32D 58151 - 58157 Lowercase alternate glyphs

U+E327 58151 g
U+E328 58152 j
U+E329 58153 p
U+E32A 58154 q
U+E32B 58155 y
U+E32C 58156 ĝ
U+E32D 58157 ĵ

I chose the list myself and the code point allocations have no official standing. Discussion is welcome. If some readers would like to use them in their own fonts, then that too is welcome.

William Overington

23 June 2008

3

I am trying to add some small capitals into my Poetry font.

I produced the artwork for 26 small capitals A..Z in a new font using the Fontstruct facility at the http://fontstruct.fontshop.com webspace, starting from a clone of the Poetry font.

I downloaded the font and processed the glyphs using FontCreator 5.6, scaling by 400% both horizontally and vertically and correcting validation errors on the artwork of the small capitals, some automatically yet three of them, namely B, G and S manually.

So, my artwork for the small capitals is correct, using the word correct in the sense that it is how I want it to be, not implying any level of artistic merit or otherwise: I am simply saying correct so that it is clear that I am not seeking to use the Glyph Transformer to produce artwork for the small capitals. I am seeking to use the Glyph Transformer to generate Private Use Area cells in which to place the artwork so that my use of small capitals in the Private Use Area follows the usage by High-Logic in the Glyph Transformer scripts and in the following document.

http://www.high-logic.com/fontcreator/Using%20Glyph%20Transformations.pdf

In learning how to add in the cells using the Glyph Transformer script I have found that small capitals for A..Z are mapped to codepoints with decimal values of 58097..58122.

I am somewhat puzzled by this as I was expecting the small capitals for A..Z to be mapped to codepoints with decimal values of 58065..58090 as 65..90 are the codepoints for A..Z whereas 97..122 are the codepoints for a..z.

Is there a reason for what appears to be the mapping of small capitals to codepoints which seem to be based on lowercase letters rather than based on capital letters?

In carrying out these experiments I have thought that it would be helpful if Tools Options… Overview Caption in FontCreator had an additional option of Unicode mappings in decimal.

William Overington

29 September 2008

4

Yes, there is. I choose the range …97 to …122 because, in a Small Caps font like Carita, which has no lowercase letter forms, the small capitals are mapped to a … z while the large capitals are mapped to A … Z.

I also reasoned that at some point in the future someone might like to add definitions in CompositeData.xml for Titling Capitals, in which case the range … 65 to … 90 would be appropriate.

In carrying out these experiments I have thought that it would be helpful if Tools Options… Overview Caption in FontCreator had an additional option of Unicode mappings in decimal.

The captions can be changed globally in Options, General to use decimal code-points. Then the decimal numbers will also be used in other dialogues such as the Insert Character or Mapping dialogues.

I have always found decimal values more useful since one can type any character in many programs by holding down Alter and typing its decimal codepoint. I have memorised hundreds of these values through frequent use, whereas I know very few in hex.

I believe that one can use hex values in CompositeData.xml, but I haven’t actually tried it. The Small Caps would be defined as $E2F1 - $E30A instead of 58097 - 58122.

It is worth knowing that in Wordpad the shortcut Alter X will convert hex values to glyphs and vice versa.

5

Excellent. I did not know of Titling Capitals in an electronic font until I read your post.

I do remember from the days of handset metal type that there were items known as titling fonts. As many readers may know, in metal type the point size is the size of the metal shank upon which the letterform is cast. So, for example, 24 point measures from above the position of the top of A to below the lowest point of the descender of a lowercase p. So setting a title using just some of the capitals of an ordinary font means that the letterforms themselves are not as big as they might be if there were not allowance for lowercase descenders. A titling font had just capitals and a few items such as some punctuation and an ampersand and letterforms occupied much more of the height on the metal body as there was no need to leave space for the descenders of g, ju, p, q and y, though allowance for the tail of Q and maybe for the lower part of J depending upon the font. With handset metal type there could be a considerable cost saving in having a titling font as larger printed letters could be obtained with a smaller financial outlay because less metal was used to make capital letterforms of a particular size. The absence of a lowercase also saved money, yet that was not a consideration as such because fonts were often sold in parts, such as capitals, lowercase and “figures and points”, so one could often buy just the capitals of any font if that was what one wanted.

I am thinking that it would be good for me to try to add some titling capitals to the Poetry font, for the experience.

Wow, so it can. I tried it with the small capitals and the decimal numbers are now displayed. Thank you.

Thank you. I tried that and it works to convert hexadecimal values to glyphs but I could not always get it to work in the other direction because if the glyph or glyphs look like a hexadecimal number it tries to produce a glyph.

William Overington

29 September 2008

6

Further to the above, I have been thinking about adding some titling capitals to the Poetry font.

I am wondering how big to make them.

The Poetry font has Win Ascent set at 2048 font units and Win Descent set at -1024 font units. Most ordinary capitals, such as H, go from 0 font units up to 1536 font units. Q goes from -384 font units to 1536 font units. M and N go from 0 font units up to 1664 font units.

The 1536 font units position is somewhat less than Win Ascent, yet that is deliberate so that there is space for accented versions of the ordinary capitals.

So from where to where, vertically, should a titling capital H go please?

If this were metal type it would be from something like the equivalent of from -512 font units up to 1536 font units, the glyphs also being wider, so that there would be larger letterforms on the same size of body.

Yet such a metal font would be free-standing, not connected with any lowercase characters. In this electronic font, the lowercase would be in the same font, hence my wondering what to do!

Readers following this thread yet who are not aware of the Poetry font may like to know that the font has its own thread in the Gallery forum, which thread includes a link to obtain a copy of the font.

http://forum.high-logic.com:9080/t/poetry/2195/1

William Overington

29 September 2008

7

If you increase the height of the Titling Capitals you may run into problems with clipping, especially if you should need them with accents. The way I would do it is reduce the stem weight, just as to produce Small and Petite Capitals I increase the weight after scaling them down. The Thin transformation should work well for this.

Compare the design of Garamond Titling with Garamond Regular Capitals on Wikipedia. The heights are the same (though it doesn’t say that they are set in the same pointsize, I assume that they are).

I might also use a slightly different design to suit the intended use. For example, when Qu is set in regular type the tail can run under the u, but with QU this may not work so well.

In practice, I think that Titling Caps would never be mixed with lowercase and would not be used at the same size, so there is no need to concern ourselves with their relative proportions. One might use, for example:

BOOK TITLE
Author

8

A comparison of the glyphs when traced into FontCreator shows that the difference is very subtle. The stroke width on the Titling Capitals (white in the illustration) is only about 10 funits less than on the Regular Capitals (black).
TitlingCapsRegularCaps.png
The outlines are rough due to the low resolution of the source image, which I enlarged 800% to make the trace.