<HTML>
<HEAD>
<!-- Created with AOLpress/2.0 -->
<!-- AP: Created on: 12-Mar-2007 -->
<!-- AP: Last modified: 28-Oct-2009 -->
<TITLE>Manipulating OpenType Lookups</TITLE>
</HEAD>
<BODY>
<BLOCKQUOTE ID="lit">
"The question is," said Humpty Dumpty, "which is to be master --" <I>(you
or the word)</I>
<P ALIGN=Right>
-- Lewis Carroll<BR>
<I>Through the Looking-Glass</I>
</BLOCKQUOTE>
<H1>
Manipulating OpenType Lookups
</H1>
<P>
<IMG SRC="fontinfo-lookups.png" WIDTH="602" HEIGHT="534" ALIGN="Right">You
use the Lookups pane of the <A HREF="fontinfo.html">Element->Font Info</A>
command to control OpenType lookups.
<P>
A lookup is a collection of commands that provide a text layout program with
a transformation to apply on the input glyph stream. A lookup might specify
how to form ligatures or how to do kerning, or any of a number of other things.
<P>
Every lookup is composed of a number of subtables. Often this complexity
is not needed and many lookups will contain just one subtable.
<P>
Every lookup is named and every subtable is named. No two lookups may have
the same name, nor may any two subtables.
<P>
There are two major classes of lookups, those that reside in the GSUB table
and substitute one glyph (or several) with another (ligatures) and those
which live in the GPOS table and position one glyph relative to another
(kerning).
<P>
The order in which the lookups and subtables appear in this dialog is important.
This is the same order in which they will be applied. All lookups in the
GSUB table will be applied before any lookups in the GPOS table.
<P>
There are buttons on the side for reordering the lookups. Up, Down, Top,
Bottom should be obvious. You can also drag and drop lookups to reorder them.
The [Sort] button will sort lookups depending on the cannonical order of
the features attached to them. This will not always be the desired ordering,
but it is a place to start.
<P>
Lookups are applied in the order listed here. Which lookups will be applied
depends on the script, language, and what features have been enabled (this
will be controlled by the text layout engine, and perhaps ultimately by the
user).
<P>
Within a lookup, the subtables will be applied in order until one of them
actually does something. Then no further subtables will be executed. Note
that this is different from the way lookups behave -- all active lookups
will always be applied, but only one subtable in a lookup will be.
<P>
Suppose for example that you have a kerning class, and a few kerning pairs.
If you put both of these into one lookup, with the kerning pairs in the first
subtable and the kerning class in the second, then if a kerning pair matches
the input stream the kerning class will not be applied (so you can have special
cases that override the data in the kerning class).
<P>
On the other hand if you put the pairs in one lookup, and the kerning class
in another lookup, then both would be applied. And the resultant motion would
be the sum of the two.
<P>
<IMG SRC="lookup-metadata.png" WIDTH="439" HEIGHT="455" ALIGN="Left">The
<CODE>[<A NAME="Add-Lookup">Add Lookup</A>]</CODE> and <CODE>[Edit
Metadata]</CODE> buttons will bring up a dialog that allows you to control
the attributes of the lookup. Every lookup has a lookup type (here "Single
Substitution") which specifies broadly what kinds of things this lookup can
do.
<DL>
<DT>
<A HREF="lookups.html#basic-subs">Single Substitution</A>
<DD>
Substitute exactly one glyph with another.
<DT>
<A HREF="lookups.html#basic-subs">Multiple Substitution</A>
<DD>
Substitutes exactly one glyph with several others.
<DT>
<A HREF="lookups.html#basic-subs">Alternate Substitution</A>
<DD>
Provides the user with a choice of substitute glyphs for each glyph.
<DT>
<A HREF="lookups.html#basic-subs">Ligature Substitution</A>
<DD>
Substitutes multiple glyphs with a single one
<DT>
<A HREF="lookups.html#contextual-subs">Contextual Substitution</A>
<DD>
Performs substitutions depending on the glyphs around the current one
<DT>
<A HREF="lookups.html#contextual-subs">Contextual Chaining Substitutions</A>
<DD>
A more complicated way of performing contextual substitutions
<DT>
<A HREF="lookups.html#contextual-subs">Reverse Contextual Chaining
Substitutions</A>
<DD>
For arabic scripts where substitutions need to be done backwards.
<DT>
<A HREF="lookups.html#sm-subs">Mac Indic State Machine</A>
<DD>
(Not an OpenType lookup at all, but one for Apple's Advanced typography tables
-- rearranges indic glyphs)
<DT>
<A HREF="lookups.html#sm-subs">Mac Contextual State Machine</A>
<DD>
(Not an OpenType lookup at all, but one for Apple's Advanced typography tables
-- contextual substitutions)
<DT>
<A HREF="lookups.html#sm-subs">Mac Insertion State Machine</A>
<DD>
(Not an OpenType lookup at all, but one for Apple's Advanced typography tables
-- contextually inserts glyphs)
<HR>
<DT>
<A HREF="lookups.html#Single-pos">Single Positioning</A>
<DD>
Moves a single glyph around
<DT>
<A HREF="lookups.html#Pair">Pair Positioning (kerning)</A>
<DD>
Moves two glyphs relative to each other
<DT>
<A HREF="lookups.html#Anchor">Cursive Positioning</A>
<DD>
For Urdu and other similar scripts. Positions one glyph at the appropriate
anchor point relative to another.
<DT>
<A HREF="lookups.html#Anchor">Mark to Base Positioning</A>
<DD>
Positions a mark (or accent) relative to a base letter
<DT>
<A HREF="lookups.html#Anchor">Mark to Ligature Positioning</A>
<DD>
Positions a mark relative to ta ligature
<DT>
<A HREF="lookups.html#Anchor">Mark to Mark Positioning</A>
<DD>
Positions a mark relative to another mark.
<DT>
<A HREF="lookups.html#contextual-pos">Contextual Positioning</A>
<DD>
Positions glyphs depending on other glyphs around them
<DT>
<A HREF="lookups.html#contextual-pos">Contextual chaining Positioning</A>
<DD>
A more complex form of the above.
<DT>
<A HREF="lookups.html#sm-kern">Mac Kerning State Machine</A>
<DD>
(Not an OpenType lookup at all, but one for Apple's Advanced typography tables
-- contextual kerning)
</DL>
<P>
A lookup may be associated with one or more feature tags each of which may
be active for various scripts and languages. You may edit the feature tag
directly, or you may click on the little box and get a list of feature tags
identified by their "friendly names" (so instead of 'smcp' you would see
"Lowercase to Small Capitals").
<P>
Similarly you may edit the script and language list directly, or you may
press on the rectangle to get a friendlier dialog (see
<A HREF="#scripts-dlg">below</A>).
<P>
Every lookup is associated with a set of flags which control its behavior.
The "Mark Class:" field is only active if there are
<A HREF="fontinfo.html#MarkClass">Mark Classes </A>defined, and the "Mark
Set:" field is active if there are <A HREF="fontinfo.html#MarkSet">Mark Sets
</A>defined. Mark classes and sets are very similar, with mark sets being
the newer (largely unsupported as I write in spring 2009) but more versatile
of the two. These provide an extension of the Ignore Marks flag -- if you
specify a mark class (or set) then all marks will be ignored except those
in the class (or set).
<P>
Every lookup must be named. You may assign the name as you wish -- except
that it must not be blank and must be unique. No two lookups may have the
same name. (If you intend to use feature files, then lookup names should
be: less than 31 characters, composed of ASCII alphanumerics, underscores
and periods -- no spaces).
<P>
For ligatures you have the option of specifying whether you want these ligatures
output in an afm file.
<P>
<IMG SRC="lang-dlg.png" WIDTH="194" HEIGHT="300" ALIGN="Right"><IMG SRC="script-lang-dlg.png"
WIDTH="439" HEIGHT="217" ALIGN="Left">The <A NAME="scripts-dlg">script</A>
dialog allows you to enter scripts and languages. Again you may edit these
directly or press on the little rectangles to get a list of friendly names.
If you choose to bring up the language dialog you may choose more than one
language in it (use the control key to make disjoint
selections).<BR Clear="Left">
<P>
After you have created a lookup you may add subtables to it. In most cases
you will only need to create one subtable, but for contextual or kerning
lookups you may need two or more (the second would contain a set of kerning
classes, while the first would contain any special cases).
<P>
The <CODE>[Add Subtable] </CODE>button will prompt you to name the new subtable,
and will then bring up a dialog to allow you to edit the commands, the data,
of that subtable. You may also bring up this dialog by selecting an existing
subtable and double clicking on it (or by pressing the <CODE>[Edit Data]</CODE>
button). The format of the dialog will depend on the lookup type.
<P>
<TABLE CELLPADDING="2">
<CAPTION>
<A NAME="basic-subs">Basic substitutions</A>
</CAPTION>
<TR VALIGN="Top">
<TD><P ALIGN=Center>
<IMG SRC="subtable-gsub-single.png" WIDTH="421" HEIGHT="269" ALT="GSUB single glyph substitution"><BR>
GSUB single glyph substitution dialog</TD>
<TD><P ALIGN=Center>
<IMG SRC="subtable-gsub-ligature.png" WIDTH="421" HEIGHT="237"><BR>
GSUB ligature substitution dialog</TD>
</TR>
</TABLE>
<P>
Many of the basic substitution dialogs are very similar. Most of the dialog
contains a list glyph names (one per line) on the left, and a list of
substitution glyph names on the right. So in the example above left, the
'smcp' feature will map the glyph named "a" to the glyph named "a.sc", "b"
to "b.sc" and so on. The multiple and alternate substitution dialogs allow
a list of replacement glyph names on the right. In a multiple substitution
the glyph will be replaced with all the glyphs named on the left, while in
an alternate substitution the word processor will provide the user with a
menu from which to pick a glyph. The ligature substitution is backwards in
that the replacement glyph is the one on the left -- if the glyphs on the
right appear together and in order then they will be replaced with the glyph
on the left. So if an "f" is followed by an "f" then it will be replaced
by an "ff" ligature glyph.
<P>
The <CODE>[Populate] </CODE>button will fill the dialog with all glyphs in
any of the scripts for which this dialog is active. If FontForge can figure
out a default replacement glyph then it will in provide that as the default
value (if it can't figure out a good replacement it will just leave the right
side blank).
<P>
The <CODE>[Add Selected] </CODE>button works in much the same way, but will
only add glyphs selected in the font view.
<P>
The <CODE>[Default Using Suffix:] </CODE>button will behave similarly (except
it will only insert entries for which FontForge can find a replacement, and
glyph name of that replacement will be found by appending the specified suffix
to the glyph name of the base glyph).
<P>
The <CODE>[Remove Empty] </CODE>button will remove any entries with no
replacement glyph(s). The <CODE>[Delete]</CODE> button will delete the currently
selected row.
<P>
The various radio buttons and check boxes at the top of the dialog control
how the base glyphs are ordered in the display -- either in
<>Alphabetic order (by glyph name) or in Unicode code point order.
If [] By Base Char is checked then composed characters like "Egrave" will
be ordered closer to "E" than to "Eth". If [] By Scripts is checked then
glyphs of one script system will be grouped together no matter how they are
ordered alphabetically or unicodally, thus "A" would be grouped with "B"
and not with "Alpha".
<H3>
<A NAME="contextual-subs">Contextual substitutions</A>
</H3>
<P>
The contextual substitution dialog is quite complicated and is discussed
in its <A HREF="contextchain.html">own section</A>. These lookups can only
live in an OpenType font, not in an Apple Advanced Typography font.
<H3>
<A NAME="sm-subs">Substitution by State Machine</A>
</H3>
<P>
The state machine substitution dialog is quite complicated and is discussed
in its <A HREF="statemachine.html">own section</A>. These lookups can only
live in an Apple Advanced Typography font.
<P>
<TABLE CELLPADDING="2">
<CAPTION>
<A NAME="Single-pos">Single Positioning</A>
</CAPTION>
<TR>
<TD><IMG SRC="subtable-gpos-singlehide.png" WIDTH="415" HEIGHT="280"></TD>
<TD><IMG SRC="subtable-gpos-singlefull.png" WIDTH="647" HEIGHT="280"></TD>
</TR>
</TABLE>
<P>
The single glyph positionioning lookup can potentially be quite complex,
but generally in any specific case very little of that complexity is used
and the dialog can be simplified. The lookup allows four different adjustments
to each glyph:
<OL>
<LI>
The glyph may be moved left or right by a certain number of em-units
<LI>
The glyph may be moved up or down by a certain number of em-units
<LI>
The glyph's horizontal advance width may be altered by a certain number of
em-units
<LI>
In fonts with vertical metrics a glyph's vertical advance may be altered
by a certain number of em-units
</OL>
<P>
In addition to this, if you have configured fontforge to support device tables,
you may provide pixel adjustments that apply to specific point sizes. At
small pixel sizes (such as those used for screen fonts) the rounding error
introduced by converting from em-units to pixels may be as large as the movement
itself. In the example at right, if the 'subs' feature is applied to the
glyph "a.sinf" that glyph should be moved downward by -560 em-units. Then
if the pixel-size of the rasterization is 10 pixels, it should be moved down
a further pixel, while if the pixel-size is 12 it should be moved up a pixel.
<P>
The [] Hide Unused Columns check box will toggle between the two views above
(well, not quite, I added two device table adjustments in the image on the
right, so the adjustment column would remain after checking the checkbox).
<P>
Often all the adjustments in a subtable will be the same (the ones here are)
so the [*] Default new entries to first check box will give all new entries
the same value as the first line.
<H3>
<A NAME="Pair">Pair</A> Positioning (kerning)
</H3>
<H4>
<IMG SRC="kerningformat.png" ALIGN="Right" WIDTH="444" HEIGHT="704">Kerning
format dialog
</H4>
<P>
When you create a kerning subtable you will first be asked whether you want
to create a kerning class subtable, or a kerning subtable with a list of
glyph pairs.
<P>
In either case, FontForge will ask if you want to it to automagically fill
up the sub-table with guesses at appropriate kerning values. This is called
"autokern"ing.
<P>
In addition, FontForge can guess appropriate glyph classes for kerning by
classes.
<P>
If you aren't interested in autokerning and want to do everything by hand,
most of this dialog is irrelevant, you are only interested in the two radio
buttons at the top and the [OK] button at the bottom.
<P>
But if you are interested in autokerning, you need to tell FontForge how
closely it should kern glyphs, and what glyphs to kern. In addition if you
want FontForge to pick kerning classes for you, you must specify the maximum
allowable amount of cumuliative error between two glyphs before they must
be in separate classes.
<P>
The <CODE>Default Separation</CODE> and <CODE>Min Kern</CODE> fields are
used in AutoKerning. The goal of kerning to to make the optical separation
between all glyphs to be constant, and the <CODE>Default Separation
</CODE>field specifies that desired value. The <CODE>Min Kern</CODE> value
is simply to prevent the dialog from filling with useless junk. If AutoKerning
suggests that two glyphs should be kerned by 1 em unit then this won't make
any difference to the human eye and there is no point in including it. So
if the kerning value (in absolute value) suggested by AutoKern is less than
<CODE>Min Kern</CODE> then fontforge will ignore that value. Selecting <CODE>[]
Touching</CODE> makes AutoKerning work in a slightly different way, instead
of trying to make the optical distance be the desired value this attempts
to make the minimum separation be the desired value (This is rarely useful,
but occasionally people want to set text where the letters actually touch
one another). The <CODE>[] Only kern glyphs closer </CODE>flag means that
FontForge will only generate negative kerning offset, which will move glyphs
closer together.
<P>
Below these fields are two panes which look rather like fontviews. Here you
may specify the selections you want to describe the glyphs being autokerned.
There are two of these panes because you usually have a slightly different
set of glyphs of interest on the left and right sides of a kerning pair.
In English, a capital letter will rarely occur in the middle of a word, so
you don't need it on the right side of a kerning pair. (Of course there are
exceptions: "FontForge" uses an internal capital, but that is rare and if
you want to save space in your font tables you can ignore it). Similarly
the closing quote character will almost never be on the left side of a kerning
pair.<BR clear="all">
<H4>
Kerning pairs
</H4>
<P>
<IMG SRC="subtable-gpos-kernpair.png" WIDTH="523" HEIGHT="537" ALIGN="Right">
The kerning class dialog is described in its
<A HREF="metricsview.html#kernclass">own section</A>, while the kerning pair
subtable dialog is described below.
<P>
Each entry in the dialog specifies a pair of glyphs and then adjustments
that may be made to each of those glyphs. Each glyph may be adjusted by any
(or all) of the four adjustments mentioned above, so there are eight potential
adjustments (and possibly device tables for each of those).
<P>
In practice kerning tends to use only one of those 8 adjustments. For left
to right text this is generally the horizontal advance of the first glyph
(x_adv#1), for right to left kerning the horizontal advance of the second
glyph is used instead, and for vertical top to bottom text the vertical advance
of the first glyph (y_adv#1). Again, the [] Hide Unused Columns may be used
to simplify the dialog.
<P>
Each time you enter a new combination, FontForge will try to guess a kerning
offset for you (autokerning). So the <CODE>Separation</CODE> text fields
have a similar meaning to that which they had above -- with the exception
that the <CODE>Min Kern</CODE> and <CODE>Only kern closer </CODE>fields will
be ignored for individual pairs.
<P>
For more elaborate usage, <CODE>[AutoKern] </CODE>button will look at all
combinations of glyphs in all scripts in the feature attached to this lookup.
The <CODE>[AutoKern Selected]</CODE> button will look at all combinations
of glyphs which are selected in the fontview (Neither of these will override
existing combinations).
<P>
Below the value matrix is a visual display of the currently selected kerning
pair. You may adjust the kerning value by moving the second glyph around.
<P>
You may select a pixel size at which to rasterize your glyphs. You may also
request to see that rasterization magnified (Note: This is different from
just rasterizing at twice the pixelsize, each pixel will be twice as big
and the effects of rounding errors will be more obvious). Magnification is
probably only useful if you are working on device tables for screen
pixelsizes.<BR CLEAR=ALL>
<H3>
<A NAME="Anchor">Anchor Positioning</A>
</H3>
<P>
<IMG SRC="subtable-gpos-anchor.png" WIDTH="421" HEIGHT="239" ALIGN="Left">Marks
(diacritical accents, vowel marks, etc.) may be attached to base glyphs using
anchors. In the latin script, the grave, accute and circumflex accents might
all attach to their base glyph at the same point, so these would all live
in one anchor class. The dot below accent attaches at a different point and
would need a different class.
<P>
You may create as many anchor classes as you like. One mark may not be in
multiple attachment classes, but a base glyph may.
<P>
Anchor classes may also be used in cursive systems such as Urdu where the
text slopes up the page. Each glyph has an entry point and an exit point,
and the entry of a glyph will be attached to the exit of the preceding glyph.
<P>
Marks may also be attached to marks. Vietnamese will often stack accents
on top of one another.
<P>
Finally, ligatures may have several attachment sites, one for each ligature
component.
<P>
<IMG SRC="anchorcontrol-base.png" WIDTH="501" HEIGHT="323" ALIGN="Right">The
above dialog may be used to create a set of classes for each type of anchor
subtable, then click on the <CODE><A HREF="anchorcontrol.html">[Anchor
Control...]</A></CODE> button to view the marks anchored to bases glyphs.
<P>
One glyph is selected (in this case a base) and its anchor point is displayed
relative to it (the blue star in the first pane after the controls). Subsequent
panes show all possible mark attachments.
<P>
You may adjust the pixelsize of the display and the magnification factor
(again the magnification factor is most useful if you are looking at small
pixelsizes.
<P>
You may provide device table adjustments for each pixelsize.<BR CLEAR=ALL>
<H3>
<A NAME="contextual-pos">Contextual Positioning</A>
</H3>
<P>
The contextual substitution dialog is quite complicated and is discussed
in its <A HREF="contextchain.html">own section</A>. These lookups can only
live in an OpenType font, not in an Apple Advanced Typography font.
<H3>
<A NAME="sm-kern">Kerning by State Machine</A>
</H3>
<P>
The state machine substitution dialog is quite complicated and is discussed
in its <A HREF="statemachine.html">own section</A>. These lookups can only
live in an Apple Advanced Typography font.
</BODY></HTML>
distrib
>
Mageia
>
5
>
i586
>
media
>
core-release
>
by-pkgid
>
6b465285c9860724a695461ac61807f4
>
files
>
544
