| Bold | Sans | Serif |
|---|---|---|
 |
 |
 |
This font is for use in e.g. embedded/FPGA/SBC projects, computer emulators, retro games.
Public Domain
- small memory footprint (512 8x8 characters occupy just 4KB)
- trivially scalable (from 8x8 to 8x16 or 16x16) in hardware or software, that is, this single font can be used at different screen resolutions, which is especially useful when the resolution is switchable
- decent support for Latin, Greek, Cyrillic and Hebrew scripts, covering approximately 50 languages
- contains a number of symbols in common use (math, punctuation, quotation, line/box-drawing)
- can be used with 8-bit code pages (e.g. IBM437, Windows-1251, ISO/IEC 8859-8) or Unicode
- for ease of porting of existing PC software and text, the first 256 chars of the font fall somewhere between code pages 437 and 850/858
- not as nice or consistent as taller fonts (e.g. 8x14 or 8x16) since there isn't much space vertically for top and bottom diacritics; a few basic characters (a, e, s, в, з, э, є) are also affected; various compromises made
- some very similarly or identically looking characters share common appearance and are thus visually indistinguishable (e.g. 3 and З, B and В)
- likewise, some less fortunate characters that don't get to have their own distinct representation in the font are approximated by the ones that do (e.g. ϊ -> ï, Ș -> Ş, Ǧ -> Ğ <- Ĝ; rationale: however wrong, it's probably better to show something recognizable and meaningful than nothing at all or a catch-all character like a box or U+FFFD <?>)
-
You can simply open one of the 512_8_*.txt files (e.g. 512_8_bold.txt) in a text editor (must be opened in UTF-8 encoding and viewed with a monospace font).
-
Here are a few pangrams prerendered with the vertically doubled font:
Style Bold 
Sans 
Serif 
-
You can install the supplied 512x*.ttf fonts in your OS and try them out in a web browser or a word processor. For typical applications you should prefer taller/vertically doubled fonts, e.g. 512x16Bold.ttf over the original non-doubled 512x8Bold.ttf.
-
You can also use the included test text files (all UTF-8 encoded), code_pages.txt and pangrams.txt to see if all of the supported font characters are rendered correctly in your project.
Currently, there are 3 styles of the font:
- bold
- noticeably partially serif due to space constraints
- most legible due to thickness and good contrast
- recommended
- sans (that is, sans serif, IOW, without serifs)
- thin, with worse contrast (text colors should be chosen carefully or adjusted for this style)
- next best after bold
- serif (most letters have serifs)
- least legible due to serifs themselves or due to their effect on the rest of the character shape
- kinda cute
- to be used sparingly and mindfully, perhaps, for artistic effect
- Substituted/approximated letters and letter-like characters:
ĈĉĊċĔĕĜĝĠĤĥĦħĨĩĬĭĴĵĿŀŌōŎŏŖŗŜŝŨũŬŭ
ƒ
ǍǎǦǧǺǻǼǽǾǿȘșȚț
ɒɔɛɜɪʃʊʌƷʒ
ίιϊ
ḂḃḊḋḞḟṀṁṖṗṠṡṪṫẀẁẂẃẄẅỲỳ - Other substituted/approximated characters:
‗’‚‼↨∟⌐▬♫✓
You can use the font's first 256 characters as if it had its own code page. The following information may be useful for that.
The font is closer to code page 850/858 than to code page 437. CP850/858 brings in additional Latin characters with diacritics that are needed in several major Latin-based languages. These "new" characters replace the less valuable line/box-drawing characters and the far incomplete set of Greek characters of CP437. This part is mostly the same in the font as in CP850/858.
However, some of CP437 characters that CP850 replaces are brought back in the font (most notably in the range 0xF0 through 0xFF).
Still, a few more characters are replaced with characters not found in either CP437 or CP850/858.
Here's an attempt to summarize the differences in the first 256 characters between this font and code pages 437 and 850:
Code Font Approxi- CP437 CP850
Char Point Index mation Index Index
☻ U+263B n/a n/a 02 02
∞ U+221E 002 EC
◘ U+25D8 n/a n/a 08 08
◙ U+25D9 n/a n/a 0A 0A
♫ U+266B ♪ 0E 0E
Š U+0160 00E
‼ U+203C ! 13 13
š U+0161 013
▬ U+25AC ─ 16 16
≠ U+2260 016
↨ U+21A8 ↕ 17 17
Ž U+017D 017
∟ U+221F └ 1C 1C
ž U+017E 01C
⌂ U+2302 07F Δ 7F 7F
------------------------------------------
₧ U+20A7 n/a n/a 9E
ƒ U+0192 f 9F 9F
ẞ U+1E9E 09F
⌐ U+2310 ┌ A9
ı U+0131 11F D5
€ U+20AC 0D5
¦ U+00A6 17E DD
▌ U+258C 008 DD
Ÿ U+0178 0DD
▐ U+2590 00A DE
¯ U+00AF n/a n/a EE
∩ U+2229 n/a n/a EF
≡ U+2261 0F0 F0
SHY U+00AD - F0
≥ U+2265 0F2 F2
‗ U+2017 _ F2
≤ U+2264 0F3 F3
¾ U+00BE 0EE F3
⌠ U+2320 1FC F4
¶ U+00B6 014 14 14,F4
ΠU+0152 0F4
⌡ U+2321 1FD F5
§ U+00A7 015 15 15,F5
œ U+0153 0F5
≈ U+2248 0F7 F7
¸ U+00B8 , F7
∙ U+2219 · F9
¨ U+00A8 n/a n/a F9
™ U+2122 0F9
√ U+221A 0FB FB
¹ U+00B9 0FF FB
ⁿ U+207F 1FB FC
³ U+00B3 0FC FC
NBSP U+00A0 " " FF FF
That is, there are several CP437 and CP850 characters completely absent from this font, without substitution/approximation. Likewise, there are some not found in either.
- Using with 8-bit code pages
You can create a look-up table to translate character codes (0 to 255) of your chosen code page (e.g. CP866) into indices into this font (0 to 511). - Using with Unicode
Use binary search or a hash table to translate each Unicode code point into an index into this font.- Binary search (needs less memory but is slower)
The generated 512_8_*.tab files list the translation key-value pairs as arguments in CDPT() and FIDX() (note that the code points for ASCII chars are not included to save space and search time). - Hash table (needs more memory but is faster)
If you prefer a hash table instead, the generated 512_8_*.hsh files contain a simple hash table implementation. The table has 256 buckets indexed by 8 least significant bits of Unicode code point. Each argument in HTVAL() contains another 7 (that is, more significant) bits of the code point (in bits 14 through 8) and the font index (in bit 15 and bits 7 through 0). This limits the supported code point values to at most U+7FFF (again the code points for ASCII chars are not included to save space and search time). Currently, at most 6 code points hash into the same bucket. Unused slots in each bucket, if any, are filled with 0xFFFF.
- Binary search (needs less memory but is slower)
- The last character in the font is special, it must be U+FFFD <?>.
- Modify e.g. 512_8_bold.txt. It should be pretty straightforward.
- Regenerate derivative files from it with the
mkfntprogram (first, compile it as shown):
$ gcc -std=c99 -O2 -Wall -Wextra -pedantic mkfnt.c -o mkfnt
$ ./mkfnt 512_8_bold.txt
Note, this will not update the *.png and *.ttf files. - If you want to make a TTF version of the font,
import datafrom e.g. 512_8_bold.bf (for square 8x8 font) or 512_8_bold.bf2 (for taller 8x16 font, that is, vertically doubled 8x8) at https://www.pentacom.jp/pentacom/bitfontmaker2/. When importing just paste the file contents after the opening brace of the tuple. Then click theBUILD FONTbutton. You'll get asked about font details. I used these on that page:- Letter Spacing: monospace, 8px
- Word Spacing: 0 px
- Metrics: 7 (or 14) ascender, 1 (or 2) descender, 0 line gap
(numbers in parentheses are for 8x16, that is, vertically doubled font)
- https://en.wikipedia.org/wiki/Code_page_437
- https://en.wikipedia.org/wiki/Code_page_850
- https://en.wikipedia.org/wiki/Mac_OS_Roman
- https://en.wikipedia.org/wiki/Western_Latin_character_sets_(computing)
- https://en.wikipedia.org/wiki/ISO/IEC_8859
- https://en.wikipedia.org/wiki/Unicode
- https://en.wikipedia.org/wiki/Plane_(Unicode)#Basic_Multilingual_Plane
- https://en.wikipedia.org/wiki/Pangram
- https://backpacker.gr/pangrams/
- https://www.pentacom.jp/pentacom/bitfontmaker2/