Recommendations for OpenType Fonts (OpenType 1.8.3)

  • Article

This chapter outlines recommendations for creating OpenType fonts.

Byte Ordering

All OpenType fonts use Motorola-style byte ordering (Big Endian).

Mixing Outline Formats

Both Microsoft and Adobe recommend against mixing outline formats within a single font. Choose the format that meets your feature requirements.

Filenames

OpenType fonts may have the extension .OTF, .TTF, .OTC or .TTC, depending on the type of outlines in the font and the desired backwards compatibility.

  • A file containing a single font resource with TrueType outlines should have either .OTF or .TTF extension. The choice between .OTF and .TTF may depend on the desire for backward compatibility on older systems or with previous versions of the font.
  • A file containing a single font resource with only CFF outline data (no TrueType outlines) should have an .OTF extension.
  • A font collection file (one that contains multiple font resources) should have either .OTC or .TTC as the extension, regardless of whether or not layout tables are present in any of the font resources, and regardless of the kind of outline data used. The .TTC extension may be used for font collection files containing font resources that use CFF outline data if needed for backward compatibility with older software that was not aware of the .OTC extension.
  • A variable font that uses OpenType Font Variations mechanisms and associated tables should use the extensions .OTF, .TTF, .OTC or .TTC following the above guidance. If there is a need to provide some indication wihtin a filename that the file contains a variable font, a recommended convention is to append “VF” (with a preceding delimiter character) at the end of the file name (before the extension) — e.g., “Selawik-VF.ttf”.

In all cases, software must determine the kind of outlines present in a font not from the filename extension but from the contents of the file.

Table Alignment and Length

All tables should be aligned to begin at offsets which are multiples of four bytes. While this is not required by the TrueType rasterizer, it does prevent ambiguous checksum calculations and greatly speeds table access on some processors.

All tables should be recorded in the table directory with their actual length. To ensure that checksums are calculated correctly, it is suggested that tables begin on 32-bit boundaries. Any extra space after a table (and before the next 32-bit boundary) should be padded with zeros.

Glyph 0: the .notdef glyph

Glyph 0 must be assigned to a .notdef glyph. The .notdef glyph is very important for providing the user feedback that a glyph is not found in the font. This glyph should not be left without an outline as the user will only see what looks like a space if a glyph is missing and not be aware of the active font’s limitation.

It is recommended that the shape of the .notdef glyph be either an empty rectangle, a rectangle with a question mark inside of it, or a rectangle with an “X”. Creative shapes, like swirls or other symbols, may not be recognized by users as indicating that a glyph is missing from the font and is not being displayed at that location.

Suggested shapes of .notdef glyph

BASE Table

The BASE table allows for different scripts in the font to specify different values for the same baseline tag. This situation could arise, for example, when a developer makes a multi-script font by combining glyphs from multiple fonts that use different baseline systems.

However, glyphs from different scripts in this font may not appear correctly aligned relative to each other when used with applications that either don’t support the BASE table or that support it but assume that a particular baseline will not vary across scripts. Furthermore, it is not always possible to determine the script of every glyph in the font, some “weakly-scripted” characters such as punctuation may be used in several scripts, and some glyphs such as ornaments may not have a script at all.

Thus, it is strongly recommended that developers construct their fonts so that all scripts in the BASE table record the same value for a particular baseline if they want their fonts to work as expected in the above situations.

If baselines vary by script, it is strongly recommended that the vendor add a DFLT script entry to the BASE table, which can be used if the script requested by the client is not matched, or if the client does not or cannot determine the script.

'cmap' Table

Windows 'cmap' Table

When building a font for Windows, a 'cmap' subtable for platform ID 3 should be included. When building a Unicode font, encoding ID 1 should be used for this subtable. (This subtable must use format 4.) When building a symbol font for Windows, encoding ID 0 should be used for this subtable.

When building a font to support Unicode supplementary characters (U+10000 to U+10FFFF)), include a 'cmap' subtable for platform ID 3, encoding ID 10. (This subtable must use format 12.) To provide compatibility with older software, a subtable for platform 3, encoding ID 1 should also be included. Depending on application support and the content of text being displayed, either the 3/1/4 or 3/10/12 subtable may be used. Therefore, glyph mappings for characters in the range U+0000 to U+FFFF must be identical between the 3/1/4 and 3/10/12 subtables. Also note that the characters mapped in the 3/10/12 subtable must be a superset of the characters mapped in the 3/1/4 subtable.

Remember that encoding records must be stored in sorted order by platform ID, then by encoding ID.

Macintosh 'cmap' Table

When building a font containing Roman characters that will be used on the Macintosh, an additional subtable is required, specifying platform ID of 1 and encoding ID of 0 (this subtable may use 'cmap' formats 0, 2, 4, or 6).

In order for the Macintosh 'cmap' table to be useful, the glyphs required for the Macintosh must have glyph indices less than 256 (since the 'cmap' subtable format 0 uses uint8 indices and therefore cannot index any glyph above 255).

The Apple 'cmap' subtable should be constructed according to Apple guidelines.

'cvt ' Table

Should be defined only if required by font instructions.

'fpgm' Table

Should be defined only if required by TrueType font instructions.

'glyf' Table

The 'glyf' table contains TrueType outline data, and can be optimized by Agfa MicroType Compression. It is recommended that developers perform this optimization prior to finalizing and adding a digital signature to the font. This is necessary for the creator’s signature to remain valid in embedded OpenType fonts.

'hdmx' Table

This table improves the performance of OpenType fonts with TrueType outlines. This table is not necessary at all unless instructions are used to control the “phantom points,” and should be omitted if bits 2 and 4 of the flags field in the 'head' table are zero. (See the 'head' table documentation in Chapter 2.) Microsoft recommends that this table be included for fonts with one or more non-linearly scaled glyphs (i.e., bit 2 or 4 of the 'head' table flags field are set).

Device records should be defined for all sizes from 8 through 14 point, and even point sizes from 16 through 24 point. However, the table requires pixel-per-em sizes, which depend on the horizontal resolution of the output device. The records in 'hdmx' should cover both 96 dpi devices (CGA, EGA, VGA) and 300 dpi devices (laser and ink jet printers).

Thus, 'hdmx' should contain entries for the following pixel sizes (PPEM): 11, 12, 13, 15, 16, 17, 19, 21, 24, 27, 29, 32, 33, 37, 42, 46, 50, 54, 58, 67, 75, 83, 92, 100. These values have been rounded to the nearest pixel. For instance, 12 points at 300 dpi would measure 37.5 pixels, but this is rounded down to 37 for this list.

This will add approximately 9,600 bytes to the font file. However, there will be a significant improvement in speed when a client requests advance widths covered by these device records.

If the font includes an LTSH table, the 'hdmx' values are not needed above the linearity threshold.

'head' Table

Although historical usage of the fontRevision value is varied, the recommended use of the field is to set it as a Fixed 16.16 value, and to report it rounded and zero-padded to three fractional decimal places. Examples: Decimal 1.5 is set as 0x00018000 and is reported as “1.500”; decimal 1.001 is set as 0x00010041 and is reported as “1.001”. All data required. If the font has been compressed with Agfa MicroType Compression, this must be indicated in the flags field of the 'head' table.

'hhea' Table

All data required. It is suggested that monospaced fonts set numberOfHMetrics to three (see 'hmtx').

'hmtx' Table

All data required. It is suggested that monospaced fonts have three entries in the numberOfHMetrics field. OpenType fonts that include CFF data must set numberOfHMetrics equal to the number of glyphs in the font and therefore cannot use the “repeat last width” optimization normally available within the 'hmtx' table.

'kern' Table

Should contain a single kerning pair subtable (format 0). Windows will not support format 2 (two-dimensional array of kern values by class); nor multiple tables (only the first format 0 table found will be used) nor coverage bits 0 through 4 (i.e. assumes horizontal data, kerning values, no cross stream, and override).

The OpenType specification allows fonts with CFF outlines to express their kerning in a 'kern' table. Many OpenType text layout engines support this. Windows GDI’s OpenType CFF driver, however, ignores the 'kern' table in a CFF OT font when it prepares kerning pairs to report via its pair kerning API.

When a 'kern' table and GPOS table are both present in a font, and an OpenType Layout engine is requested to apply kerning to a run of text of a particular script and language system: (a) If the number of 'kern' feature lookups in the resolved language system in the GPOS table is zero, then the 'kern' table should be applied, followed by any remaining GPOS features requested. (b) If the number of 'kern' feature lookups in the resolved language system in the GPOS table is non-zero, then all GPOS lookups, including the 'kern' lookups, should be applied in the usual way and the 'kern' table data ignored.

If a 'kern' table present but no GPOS table is present in the font, then an OpenType Layout engine should apply the 'kern' table to the text, regardless of the resolved language system of the text.

If compatibility with legacy environments is not a concern, font vendors are encouraged to record kerning in the GPOS table’s 'kern' feature and not in the 'kern' table.

OpenType Font Variations mechanisms do not include any way to represent variation of data within a 'kern' table. Therefore, kerning in a variable should be implemented using the GPOS table.

'loca' Table

All data required for fonts with TrueType outlines. We recommend that local offsets should be 16-bit aligned, in both the short and long formats of this table.

The actual ordering of the glyphs in the font can be optimized based on expected utilization, with the most frequently used glyphs appearing at the beginning of the font file. Additionally, glyphs that are often used together should be grouped together in the file. The will help to minimize the amount of swapping required when the font is loaded into memory.

LTSH Table

This table improves the performance of OpenType fonts with TrueType outlines. The table should be used if bit 2 or 4 of flags in 'head' is set.

'maxp' Table

All data required for a font with TrueType outlines. Fonts with CFF or CFF2 data must only fill the numGlyphs field.

'name' Table

Platform and encoding IDs in the 'name' table should be consistent with those in the 'cmap' table. If they are not, the font will not load in Windows. When building a Unicode font for Windows, the platform ID should be 3 and the encoding ID should be 1. When building a symbol font for Windows, the platform ID should be 3 and the encoding ID should be 0, and the referenced string data must be encoded in UTF-16.

When building a font containing Roman characters that will be used on the Macintosh, an additional name record is required, specifying platform ID of 1 and encoding ID of 0.

Each set of name records should appear for US English (language ID = 0x0409 for Microsoft records, language ID = 0 for Macintosh records); additional language strings for the Microsoft set of records (platform ID 3) may be added at the discretion of the font vendor.

Remember that, despite references to “first” and “second,” the name record must be stored in sorted order (by platform ID, encoding ID, language ID, name ID). The 'name' table platform/encoding IDs must match the 'cmap' table platform/encoding IDs, which is how Windows knows which name set to use.

Name strings

We recommend using name IDs 8–12, to identify manufacturer, designer, description, URL of the vendor, and URL of the designer. URLs must contain the protocol of the site for example, http// or mailto: or ftp://. The OpenType font properties extension can enumerate this information to the users.

The Subfamily string in the 'name' table should be used for variants of weight (ultra light to extra black) and style (oblique/italic or not). So, for example, the full font name of “Helvetica Narrow Italic” should be defined as Family name “Helvetica Narrow” and Subfamily “Italic”. This is so that Windows can group the standard four weights of a font in a reasonable fashion for non-typographically aware applications which only support combinations of “bold” and “italic.”

The Full font name string usually contains a concatenation of strings 1 and 2. If the font is “Regular” as indicated in string 2, then sometimes only the family name contained in string 1 is used for the full font name. In many contexts, the full font name is what will be exposed to users.

In variable fonts, the Typographic Family and Typographic Subfamily names (name IDs 16 and 17) are required. Applications that support OpenType Font Variations will typically present to users the Typographic Family name along with the Typographic Subfamily name or alternative subfamily names for named instances, as specified in the font variations ('fvar') table. In some situations, a font vendor may want to make available a variable font as well as some set of non-variable fonts corresponding to the named instances of the variable font. In such situations, the vendor may want to have distinct family names for the family implemented as a variable font and the family implemented using several non-variable fonts. In that case, a suggested convention is to append “VF” at the end of the variable-font family name. Note, however, that this will result in distinct families, and content formatted with the one may not display as intended in some contexts if only the other is available. For example, if a document is formatted using the regular and bold instances of a variable font with family name “Selawik VF” and then the document is viewed in a context in which only the non-variable fonts Selawik Regular and Selawik Bold are available, the viewing application will generally not be able to associate the non-variable fonts available to it with the formatting declarations in the content.

OpenType fonts that include a name with name ID of 6 should include these two names with name ID 6, and characteristics as follows:

  1. Platform: 1 [Macintosh]; Platform-specific encoding: 0 [Roman]; Language: 0 [English].
  2. Platform: 3 [Windows]; Platform-specific encoding: 1 [Unicode]; Language: 0x409 [English (American)].

Names with name ID 6 other than the above two, if present, may be ignored.

When translated to ASCII, these two name strings must be identical and restricted to the printable ASCII subset, codes 33 through 126, except for the 10 characters: '[', ']', '(', ')', '{', '}', '<', '>', '/', '%'. Some implementations have a 63-character length limit; however, a 127-character length limit is recommended.

The term “PostScript Name” here means a string identical to the two identical name ID 6 strings described above.

Depending on the particular font format that PostScript language font uses, the invocation method for the PostScript font differs, and the semantics of the resulting PostScript language font differ. The method used to invoke this font depends on the presence of Name ID 20.

If a Name ID 20 is present in this font, then the default assumption should be that the PostScript Name defined by name ID 6 should be used with the PostScript 'composefont' invocation. This PostScript Name is then the name of a PostScript language CIDFont resource which corresponds to the glyphs of the OpenType font. This name is valid to pass, with an appropriate PostScript language CMap reference, and an instance name, to the PostScript language 'composefont' operator.

If no Name ID 20 is present in this font, then the default assumption should be that the PostScript Name defined by name ID 6 should be used with the PostScript 'findfont' invocation, for locating the font in the context of a PostScript interpreter. This PostScript Name is then the name of a PostScript language Font resource which corresponds to the glyphs of the OpenType font. This name is valid to pass to the PostScript language 'findfont' operator. This does not necessarily imply that the resulting font dictionary accepts an /Encoding array, such as when the font referenced is a Type 0 PostScript font.

This specification applies only to data fork OpenType fonts. Macintosh resource-fork TrueType and other Macintosh sfnt-wrapped fonts supply the PostScript font name to be used with the 'findfont' invocation, in order to invoke the font in a PostScript interpreter, in the FOND resource style-mapping table.

Developers may choose to ignore the default usage when appropriate. For example, PostScript printers whose version is earlier than 2015 cannot process CID font resources and a CJK OpenType/CFF-CID font can be downloaded only as a set of Type 1 PostScript fonts. Legacy CJK TrueType fonts, which do not have a Name ID 20, may still be most effectively downloaded as a CID font resource. Definition of the full set of situations in which the defaults should not be followed is outside the scope of this document.

The value held in the name ID 20 string is interpreted as a PostScript font name that is meant to be used with the 'findfont' invocation, in order to invoke the font in a PostScript interpreter.

If the name ID 20 is present in a font, there must be one name ID 20 record for every Macintosh platform 'cmap' subtable in that font. A particular name ID 20 record is associated with the encoding specified by the matching 'cmap' subtable. A name ID 20 record is matched to a 'cmap' subtable when they have the same platform and platform-specific encoding IDs, and corresponding language/version IDs. Name ID 20 records are meant to be used only with Macintosh 'cmap' subtables. The version field for a 'cmap' subtable is one more than the language ID value for the corresponding name ID 20 record, with the exception of the 'cmap' subtable version field 0. This version field, meaning “not language-specific”, corresponds to the language ID value 0xFFFF, or decimal 65535, for the corresponding name ID 20 record.

When translated to ASCII, this name string must be restricted to the printable ASCII subset, codes 33 through 126, except for the 10 characters: '[', ']', '(', ')', '{', '}', '<', '>', '/', '%'.

This specification applies only to data fork OpenType fonts. Macintosh resource-fork TrueType and other Macintosh sfnt-wrapped fonts supply the PostScript font name to be used with the 'findfont' invocation, in order to invoke the font in a PostScript interpreter, in the FOND resource style-mapping table.

A particular Name ID 20 string always corresponds to a particular Macintosh 'cmap' subtable. However, some host OpenType/TTF fonts also contain a 'post' table, format 4, which provides a mapping from glyph ID to encoding value, and also corresponds to a particular Macintosh 'cmap' subtable. Unfortunately, the 'post' table format 4 contains no provision for identifying which Macintosh 'cmap' subtable it matches, nor for providing more than one mapping. Host fonts which contain a 'post' table format 4, should also contain only a single Macintosh 'cmap' subtable, and a single Name ID 20 string. In the case where there is more than one Macintosh 'cmap' subtable and more than one Name ID 20 string, there is no definition of which one matches the 'post' table format 4.

OS/2 Table

All data required.

PANOSE values

PANOSE values may improve the user’s experience for font selection in some applications or font management utilities.

If the font is a symbol font, the first byte of the PANOSE value must be set to “Latin Pictorial” (value = 5). The PANOSE evaluation document is on-line at https://monotype.github.io/panose/.

In a variable font that uses OpenType Font Variation mechanisms, there is no way to represent different PANOSE values for different instances supported by the font. The PANOSE values can be set based on the default instance.

sTypoAscender, sTypoDescender and sTypoLineGap

The sTypoAscender, sTypoDescender and sTypoLineGap fields are used to specify the recommended default line spacing for single-spaced horizontal text. The baseline-to-baseline distance is calculated as follows:

OS/2.sTypoAscender - OS/2.sTypoDescender + OS/2.sTypoLineGap

sTypoAscender should be used to determine an optimal default offset from the top of a text frame to the first baseline. Similarly, sTypoDescender should be used to determine an offset from the last baseline to the bottom of the text frame.

It is often appropriate to set the sTypoAscender and sTypoDescender values such that the distance (sTypoAscender - sTypoDescender) is equal to one em. This is not a requirement, however, and may not be suitable in some situations. For example, if a font is designed for a script that (in horizontal layout) requires greater vertical extent relative to Latin script but also needs to support Latin script, and needs to have the visual size of Latin glyphs be similar to other fonts when set at the same text size, then the (sTypoAscender - sTypoDescender) distance for that font would likely need to be greater than one em.

The sTypoLineGap value will often be set such that the default baseline-to-baseline distance is approximately 120% of the em. For example, in the Minion Pro font family, fonts are designed on a 1000 units-per-em grid, the (sTypoAscender - sTypoDescender) distance is one em, and the sTypoLineGap value is set to 200.

In CJK (Chinese, Japanese, and Korean) fonts, it is permissible for the sTypoDescender and sTypoAscender fields to specify metrics different from the HorizAxis.ideo and HorizAxis.idtp baselines in the BASE table. However, some applications may not read the BASE table at all but simply use the sTypoDescender and sTypoAscender fields to describe the bottom and top edges of the ideographic em-box. If developers want their fonts to work correctly with such applications, they should ensure that any ideographic em-box values in the BASE table describe the same bottom and top edges as the sTypoDescender and sTypoAscender fields. See OpenType CJK Font Guidelines and Ideographic Em-Box for more details.

usLowerOpticalPointSize and usUpperOpticalPointSize

Use of the usLowerOpticalPointSize and usUpperOpticalPointSize fields has been superseded by the STAT table. See Families with Optical Size Variants, below, for more information.

'post' Table

All information required, although the VM Usage fields may be set to zero. OpenType fonts containing CFF outlines use only format 3.0 of the 'post' table. Glyph names must be assigned as described in the Adobe Glyph List Specification, which specifies glyph naming conventions for all Unicode characters as well as those that don’t have standard Unicode values such as certain ligatures or glyphic variants.

Note that names for all glyphs must be supplied as it cannot be assumed that all Microsoft platforms will support the default names supplied on the Macintosh.

NOTE: The PostScript glyph name must be no longer than 31 characters, include only uppercase or lowercase English letters, European digits, the period or the underscore, i.e. from the set [A-Za-z0-9_.] and should start with a letter, except the special glyph name “.notdef” which starts with a period.

'prep' Table

Should be defined only if required by the TrueType font instructions.

VDMX Table

This table improves the performance of OpenType fonts with TrueType outlines. It should be present if hints cause the font to scale non-linearly. If not present, the font is assumed to scale linearly. Clipping may occur if values in this table are absent and font exceeds linear height.

General Recommendations

TrueType Collections

The process of building TTC files involves paying close attention to the issue of glyph renumbering in a font and the side effects that can result, in the 'cmap' table and elsewhere. The fonts to be merged must also have compatible TrueType instructions — i.e. their pre-programs, function definitions, and control values must not conflict.

Optimized Table Ordering

OpenType fonts with TrueType outlines are more efficient in the Windows operating system when the tables are ordered as follows (from first to last):

'head', 'hhea', 'maxp', OS/2, 'hmtx', LTSH, VDMX, 'hdmx', 'cmap', 'fpgm', 'prep', 'cvt ', 'loca', 'glyf', 'kern', 'name', 'post', 'gasp', PCLT, DSIG

The initial loading of an OpenType font containing CFF data will be more efficiently handled if the following sfnt table ordering is used within the body of the sfnt (listed from first to last):

'head', 'hhea', 'maxp', OS/2, 'name', 'cmap', 'post', 'CFF ', (other tables, as convenient)

Non-Standard (Symbol) Fonts

Non-standard fonts such as Symbol or Wingdings(tm) have special requirements for Microsoft platforms. These requirements affect the 'cmap', 'name' and OS/2 tables; the requirements and recommendations for all other tables remain the same.

For the Macintosh, non-standard fonts can continue to use platform ID 1 (Macintosh) and encoding ID 0 (Roman character set). The 'cmap' subtable should use format 0 and follow the standard PostScript character encodings.

For non-standard fonts on Microsoft platforms, however, the 'cmap' and 'name' tables must use platform ID 3 (Microsoft) and encoding ID 0 (Unicode, non-standard character set). Remember that 'name' table encodings should agree with the 'cmap' table. Additionally, the first byte of the PANOSE value in the OS/2 table must be set to ‘Latin Pictorial’ (value = 5).

The 'cmap' subtable (platform 3, encoding 0) must use format 4. The character codes should start at 0xF000, which is in the Private Use Area of Unicode. It is suggested to derive the format 4 encodings by simply adding 0xF000 to the format 0 (Macintosh) encodings.

Under Windows, only the first 224 characters of non-standard fonts will be accessible: a space and up to 223 printing characters. It does not matter where in user space these start, but 0xF020 is suggested. The usFirstCharIndex and usLastCharIndex values in the OS/2 table would be set based on the actual minimum and maximum character indices used.

Baseline to Baseline Distances

The OS/2 table fields sTypoAscender, sTypoDescender, and sTypoLineGap free applications from Macintosh- or Windows-specific metrics which are constrained by backward compatibility requirements. The following discussion only pertains to the platform-specific metrics.

The suggested Baseline to Baseline Distance (BTBD) is computed differently for Windows and the Macintosh, and it is based on different OpenType metrics. However, if the recommendations below are followed, the BTBD will be the same for both Windows and the Mac.

Windows

The Windows metrics in the table below are returned as part of the logical font data structure by the GDI CreateLogFont( ) API.

Windows Metric OpenType Metric
ascent usWinAscent
descent usWinDescent
internal leading usWinAscent + usWinDescent - unitsPerEm
external leading MAX(0, LineGap - ((usWinAscent + usWinDescent) - (Ascender - Descender)))

The suggested BTBD = ascent + descent + external leading

It should be clear that the “external leading” can never be less than zero. Pixels above the ascent or below the descent will be clipped from the character; this is true for all output devices.

The usWinAscent and usWinDescent are values from the OS/2 table. The unitsPerEm value is from the 'head' table. The LineGap, Ascender and Descender values are from the 'hhea' table.

Macintosh

Ascender and Descender are metrics defined by Apple and are not to be confused with the Windows ascent or descent, nor should they be confused with the true typographic ascender and descender that are found in AFM files. The Macintosh metrics below are returned by the Apple Advanced Typography (AAT) GetFontInfo( ) API.

Macintosh Metric OpenType Metric
ascender Ascender
descender Descender
leading LineGap

The suggested BTBD = ascent + descent + leading

If pixels extend above the ascent or below the descent, the character will be squashed in the vertical direction so that all pixels fit within these limitations; this is true for screen display only.

Making Them Match

If you perform some simple algebra, you will see that the suggested BTBD across both Macintosh and Windows will be identical if and only if:

LineGap >= (yMax - yMin) - (Ascender - Descender)

Style Bits

For backwards compatibility with previous versions of Windows, the macStyle bits in the 'head' table will be used to determine whether or not a font is regular, bold or italic (in the absence of an OS/2 table). This is completely independent of the usWeightClass and PANOSE information in the OS/2 table, the ItalicAngle in the 'post' table, and all other related metrics. If the OS/2 table is present, then the fsSelection bits are used to determine this information.

Drop-out Control

Drop-out control is needed if there is a difference in bitmaps with dropout control on and off. Two cases where drop-out control is needed are when the font is rotated or when the size of the font is at or below 8 ppem. Do not use SCANCTRL unless needed. SCANCTRL or the drop-out control rasterizer should be avoided for Roman fonts above 8 points per em (ppem) when the font is not under rotation. SCANCTRL should not be used for “stretched” fonts (e.g. fonts displayed at non-square aspect ratios, like that found on an EGA).

Embedded Bitmaps

Three tables are used to embed bitmaps in OpenType fonts. They are the EBLC table for embedded bitmap locators, the EBDT table for embedded bitmap data, and the EBSC table for embedded bitmap scaling information. OpenType embedded bitmaps are also called “sbits”.

The behavior of sbits within an OpenType font is essentially transparent to the client. A client need not be aware whether the bitmap returned by the rasterizer comes from an sbit or from a scan-converted outline.

The metrics in 'sbit' tables overrule the outline metrics at all sizes where sbits are defined. Fonts with 'hdmx' tables should correct those tables with 'sbit' values.

“Sbit-only” fonts, that is fonts with embedded bitmaps but without outline data, are permitted. Care must be taken to ensure that all required OpenType tables except 'glyf' and 'loca' are present in such a font. Obviously, such fonts will only be able to return glyphs and sizes for which sbits are defined.

  1. These metrics are returned as part of the logical font data structure by the GDI CreateLogFont() API.
  2. These metrics are returned by the Apple Advanced Typography (AAT) GetFontInfo() API.

OpenType CJK Font Guidelines

This section provides a checklist of links to various CJK-related sections of the OpenType specification. Some items are requirements; others, recommendations:

  1. The ideographic em-box of an OpenType font will be determined as described in the section Ideographic Em-Box in the Baseline Tags section of the OpenType Layout Tag Registry. Also see the description for OS/2.sTypoAscender and OS/2.sTypoDescender, and the BASE table recommendation section above.
  2. CJK font vendors can choose to provide the ideographic character face (ICF) metrics, which applications can use for accurate text alignment. This is described in the section Ideographic Character Face in the Baseline Tags section of the OpenType Layout Tag Registry.
  3. All OpenType fonts that are used for vertical writing must include a Vertical Header ('vhea') table and a Vertical Metrics ('vmtx') table. It is strongly recommended that CFF OpenType fonts that are used for vertical writing include a Vertical Origin (VORG) table.
  4. If an OpenType font with CFF outlines is to be used for vertical layout, some legacy platforms may require that a Vertical Rotation ('vrt2') feature be present in the Glyph Substitution (GSUB) table. See the Feature Tags section of the OpenType Layout Tag Registry for a description of and further requirements for this feature.
  5. See the Feature Tags section of the OpenType Layout Tag Registry for descriptions of currently registered OpenType layout features, such as Alternate Half Widths ('halt') and Traditional Forms ('trad'), that can be specified in the font.

Stroke Reduction in Variable Fonts

When designing a font family to support a number of variations, there may be cases in which it is desirable to make significant, structural changes to particular glyphs for certain variations. A common example is stroke reduction for heavier weights or narrower widths, simplifying the structure of a glyph so that counters do not become filled and disappear at smaller text sizes. Within a variable font, two techniques might be used to implement a stroke-reduction effect:

  • Use a pair of slightly-overlapping intermediate regions within the variation data for a glyph in order to introduce deltas for particular contour points that result in the desired structural change and that apply for only problem ranges on one or more axes.
  • Use an OpenType Layout Required Variation Alternates feature in combination with a FeatureVariations table within the GSUB table to perform a glyph substitution when a variation instance is selected in some range along one or more axes.

While both techniques are possible, it should be noted that the first technique, using overlapping intermediate-regions, can be tricky to implement and may result in unexpected or undesired results if an instance is selected using arbitrary axis values in the range over which the transition occurs. The second technique is recommended as it will generally be easier to implement and maintain, and provides the font designer better control over behavior near the point of transition.

Families with Optical Size Variants

In families that have fonts for different optical sizes or in variable fonts that support the optical size ('opsz') design axis, a STAT table with format 2 axis value tables should be used to indicate text size ranges for which the different optical-size variants or variable-font named instances are recommended. This supersedes the use of the usLowerOpticalPointSize and usUpperOpticalPointSize fields in the OS/2 table, and the OpenType Layout 'size' feature.

When creating an axis value table to correspond to the font or named instance that is intended for the largest text sizes, the upper text size limit should be, effectively, infinity. To represent this in a format 2 axis value table, set the rangeMaxValue to 0x7FFFFFFF.