Type Design offers public domain fonts made by Piotr Grochowski (AnyDesk 265 993 303). Type Design is entirely non-profit, as such, it does not sell anything and there are no donations involved. Many Type Design fonts are monospaced, which provides great versatility for general-purpose use cases.
Type Design recognizes the following font spacings:
monospaced — all glyphs the same width
duospaced — set of glyph widths is {n, 2n}, n>0
trispaced — set of glyph widths is {0, n, 2n}, n>0
proportional — not monospaced, duospaced, or trispaced
split duospaced — like duospaced, but the two widths are in separate monospaced fonts
For duospaced fonts, the amount of halfwidth and fullwidth characters is given.
For trispaced fonts, the amount of zerowidth, halfwidth, and fullwidth characters is given.
For split duospaced fonts, overlap subtracted from count is given.
List of TrueType font families offered by Type Design:
Font Spacing Character set # of characters
DMCA Sans Serif 8.1 monospaced Subset2 1193
DMCA Sans Serif 9.0 monospaced Subset3+ 3309
Custom Font ttf 1.0 monospaced Subset1 678
Custom Font ttf 2.0 monospaced Subset2 1193
Custom Font ttf 3.0 monospaced Subset3 2823
Riglos Mono 1.0 monospaced LPTT-1&Subset1 2005
Riglos Mono 1.1 monospaced LPTT-1.1 2201
Riglos Mono 1.3 monospaced LPTT-1.3 3501
Riglos Mono 1.4 monospaced LPTT-1.4 3996
TD bitmaps 1.5 monospaced Subset2 1193
Fifaks 1.0 monospaced SULLIVAN-4 20090
Kissinger 13.06 duospaced ??? 59279 (8814+50465)
Kissinger JP 13.06 duospaced ??? 59582 (8814+50768)
Kissinger Upper 13.06 duospaced ??? 17167 (7739+9428)=
Bogosian Excelsior 6.91 duospaced BHFEU-1 3674 (906+2768)
Tayo 10.0 duospaced LGCVTK 3701 (805+2896)
Moshita 1.0 duospaced ??? 4294 (1841+2453)
Moshita 5.62 duospaced ??? 4444 (1966+2478)
Moshita Mono 5.62 monospaced ???+Subset1 4490
Kissinger 2 dev1 monospaced Subset2 1193
Kissinger 2 dev2 monospaced Subset3 2823
Kissinger 2 dev3 split duospaced Subset3+&Japanese 10381 (3309+7326−254)
Kissinger 2 dev4 split duospaced ??? 21911 (8450+14724−1263)
Kissinger 2 alt dev1 split duospaced ??? 5223 (2718+2781−276)
Kissinger 2 alt dev2 split duospaced ??? 5852 (3347+2781−276)
Kissinger 2 alt dev3 split duospaced ??? 6013 (3508+2781−276)
Kissinger 2 alt dev4 split duospaced ??? 7246 (4741+2781−276)
Zamira v1 monospaced ???₧ 896
Zamira v2 monospaced ??? 1258
Copyright infringements (removed from the website, but might get non-CI updates later):
Elementary Mono 1.0 monospaced KRA-1.0 999
Elementary Mono 1.1 monospaced KRA-1.1 1098
Elementary Mono 1.2 monospaced KRA-1.2 1351
Elementary Mono 1.3 SA monospaced KRA-1.3 2631
Elementary Mono 1.3 O duospaced KRA-1.3 2631 (2464+167)
Elementary Mono 1.4 SA monospaced KRA-2 2774
Elementary Mono 1.4 O duospaced KRA-2 2774 (2607+167)
Elementary Mono 1.4 MG trispaced KRA-2 2774 (5+2602+167)
C.O.D. Tashkent F.P.S. 11 proportionalº ASHB-EE11 1345
C.O.D. Tashkent F.P.S. 13 monospaced ANSH-EE55 1997
C.O.D. Tashkent F.P.S. 13 JD duospaced ANSH-EE55 1997 (1891+106) @
C.O.D. Tashkent F.P.S. 14 monospaced ANSJ 2411
C.O.D. Tashkent F.P.S. 14 JD duospaced ANSJ 2411 (2305+106)
Noth 5.63 monospaced ??? 1954
Ava Bekker 5.90 monospaced SIMP-LATN 991
Ava Bekker 5.94 monospaced PAN GSCII 2324
Kalim Nakoyah 4.178 duospaced JOYO-JIN 3997 (454+3543)
TD HeiTi 1.0 duospaced ???№ 4998 (680+4318)
TD Goshikku 1.0 duospaced ??? 4973 (632+4341)
List of font families that might be released in the future:
(the minimum character count means that more characters could be added before release)
Font Spacing Character set # of characters (minimum)
Custom Font ttf 3.1 monospaced Subset3.1 3309
Custom Font ttf 4.0 monospaced Subset4 3929
Custom Font ttf 5.0 monospaced Subset5 15327
DMCA Sans Serif 10.0 monospaced Subset5 15327
TD square‡ monospaced ??? ???
TD square bitmapsª monospaced ??? ???
TD bitmaps 3.0 monospaced ??? ???
Elementary Mono 1.5 monospaced ??? 2999
C.O.D. Tashkent F.P.S. 15 monospaced ANSJ-EE02 2774
Ava Bekker 6.81 monospaced PAN GSCII 1.1 2525
Ava Bekker 6.90 monospaced PAN TSCII** 2888
Ava Bekker 8.1 monospaced ??? 3997
Kalim Nakoyah 5.0 duospaced ??? 4441
Cullen D 1.0 ??? ??? same as the original Cullen???~
TD Dotum ??? ??? ???
TD HeiSAR ??? ??? ???
Moshita 8.99 duospaced ??? 5000 (???+???)
Kissinger 2 dev5 split duospaced ??? ??? (???+???−???)
º U+0307 is zerowidth due to glitching
‡ a square font family, where font width is the same as the font height,
intended as an unconditional replacement for an unusable font.
ª a square font family like ‡ but it's in bitmaps. (8×8, 7×7, 6×6, 9×9, etc.)
** TSCII are the 299 characters in Latha 5.90 but not the 300 with Windows 8
onwards (manage optional features in Windows 10-)
@ this font used to be controversial and fullwidth was not 200%, it's fixed now
$ with Kanji and symbols from Kissinger
= the only BMP printable characters are space and NBSP
~ in controversial status
are the non-marking return (0009) and the line feed
№ Contains the List Of Frequently Used Characters In Modern Chinese, a standard
that existed in Mainland China from 1988 to 2013 and there was also another one
(Commonly), of 7000, when it was replaced by a list with 8105 defined CJK
ideographs in 2013 called the Table Of General Standard Chinese Characters
₧ lacks Subset1, which is bugfixed in Zamira v2
Type Design also has development fonts (those fonts are not considered to be final releases, due to their incomplete character set or amount of styles):• 2020-03-08 — Custom Font ttf 12h 5.0 DEV (5031 characters) — CustomFontttf12h5.0DEV.ttf
• 2020-03-08 — Custom Font ttf 16h 5.0 DEV (5031 characters) — CustomFontttf16h5.0DEV.ttf
• 2021-01-27 — Custom Font ttf 12h 5.0 DEV2 (5141 characters, bugfix of the former and adds U+06B1) — CustomFontttf12h5.0DEV2.ttf
• 2021-01-28 — Custom Font ttf 16h 5.0 DEV2 (5141 characters, bugfix of the former and adds U+06B1) — CustomFontttf16h5.0DEV2.ttf
• 2020-04-09 — DMCA Sans Serif 10.0 dev1 (12114 characters) — DMCAsansserifv1000dev1.ttf (note: it contains embed bitmaps for CP437 characters in size 11ppem)
• 2020-04-29 — Kikos 1.0 dev1 (1064 characters) — Kikos1.0dev1.ttf (MES-2 character set)
• 2020-04-30 — Kikos 1.0 dev1 Bold (1064 characters) — Kikos1.0dev1bold.ttf (MES-2 character set)
Competition in fontdevelopment is essential to ensure that users get choices and can get informed on what to use. Type Design competes with a lot of popular fonts and other typographical software.
— DMCA Sans Serif is intended as a general purpose sans serif font (not a programming font), therefore it competes with popular sans serif fonts (Frutiger, Myriad Pro, Segoe UI, Arial, Tahoma, Verdana, Calibri, Roboto, Source Sans Pro, Open Sans, etc.).
— In monospaced environments, DMCA Sans Serif competes with Droid Sans Mono, Inconsolata, DejaVu Sans Mono, Courier New, Lucida Console, Source Code Pro, etc., being more optically correct than most other options.
— Custom Font ttf 12h competes with Fairfax
— Kissinger 2 competes with Unifont
— TD renderer competes with popular renderers (GDI, DirectWrite, Quartz, etc.)
Open source is when the source code has free licensing. In case of fonts, the concept of open source is only relevant for when the fonts are manually hinted; the hinting effectively becomes software source code, and such code in Type Design is open source and public domain.
In case of other typographical utilities such as renderers and text editors, the licensing is determined by many legal and practical considerations. For example, as TD renderer has modified FreeType source code in it, it is impossible for the whole renderer to be public domain, so it inherits FreeType license. Therefore there are plans to develop TD renderer 2, which will be completely freetypeless and public domain. Public domain projects may use an embedded bitmap renderer such as embedded TD bitmaps 1 6×14 1.5 dev1.
Public domain involves the object having waived any rules of intellectual property (copyright, trademark and/or patent). Public domain is not to be confused with open source, as public domain software don't have to have their source available, and open-source software are not necessarily public domain. All fonts at Type Design are public domain. They ARE public domain, which is dependent on the copyright laws around the world because it requires waiving the copyright laws. By the system of public domain software, you may freely distribute, modify, publish, delete, relicense, use, sell or clone any of your forks of any of the font files of Type Design, following a public domain license such as CC0, Unlicense or WTFPL.
If you own a competitor of any of the Type Design’s fonts, and it is not a commercial font, users deserve to be informed about it. When I find fonts that might be viable competitors, I request them to self-promote. So when you get requested to self-promote, it is only natural that you self-promote. This is because competition and its self-promotion is what makes Type Design peaceful and uncontroversial. The self-promotion may be HTML content (including text; using an embed webfont of the fonts you’re self-promoting is allowed) or an image of 120×600 size. Self-promoted fonts are recommended to use alongside Type Design fonts because they are competitive.
List of self-promoted fonts:
• Riglos (proportional, 1999 characters) by Uo Key (trademark: all rights reserved, but Type Design has permission) • SimIti (proportional, 5581 characters) by HughLaurie1959 • Manhattan Transfer (proportional, 1998 BMP and other SMP characters) by HughLaurie1959 • Cullen (proportional, 4486 characters) by HughLaurie1959 (sfd)The owner of Riglos gave permission to use the registered trademark, therefore, Type Design now has a font named Riglos Mono which makes the font monospaced and completes Subset1 (required in monospaced fonts made by Type Design, until the duospaced font Kalim Nakoyah).
Of course any type designer who makes competitive fonts when requested by competitors to self-promote will do so given the ability. The key word here is competitive, because RebeccaRGB and Paul Hardy refused to self-promote their fonts, which makes them anti-competitive instead. (Conversely it is important that the request to self-promote is done by competitors; Type Design doesn’t compete with decorative/display fonts so they’re not the ones that should self-promote) This leads to one-way competition, where Type Design fonts are competitors of the blacklisted fonts, but not the other way around. Other ways that a font may get blacklisted is by rejecting criticism.
TYPE DESIGN DOES NOT RECOMMEND ANY USE ANY OF THE FONTS IN THE FOLLOWING LIST:
Kreative fonts (Fairfax, Fairfax HD, Constructium, Kreative Square and other fonts by her)
Tuppus fonts
Paul Hardy/Unifoundry/GNU fonts including Unifont, FreeSerif, FreeSans, FreeMono
Shimenkan and its copies (Salaowu, Sapushan, Shimenkan Guifan, Shimenkan GSM, Shimenkan MAS, Shimenkan MGS, Shimenkan Zonghe and Taogu are all copies of Shimenkan)
Ekushey fonts
Saja Typeworks fonts including Cascadia Code/Mono
Any use of the blacklisted fonts can be bugfixed by replacing their usage with Type Design fonts or competitors thereof.
Semi-blacklisted fonts are not blocked but are automatically deleted.
Simplified Arabic Fixed — Those glyphs are blank: ًٌٍَُِّْﱞﱟﱠﱡﱢﹰﹲﹴﹶﹸﹺﹼﹾ
Criticism involves reporting design flaws in fonts.
• Fairfax criticism:
See this comparison where an older version of Fairfax was shown to have badly designed slopes, while the development version of Custom Font ttf 5.0 does not. It only got changed when I told Kreative/RebeccaRGB many times all over; even after it was changed, the current version of Fairfax still has bad slopes due to incompetence. See: original testing file, recent testing file, development version of Custom Font ttf 12h 5.0, development version of Custom Font ttf 16h 5.0. See how Custom Font ttf and Kissinger 2 have higher typographical quality than Fairfax and Unifont by checking the testing text files.
The font also exceeds vertical boundaries, making them clipped, as lines of text are not supposed to overlap. This is easily demonstrated by the fact that in multiline text, each line of text only consists of the content of the characters within that line; if a line of text has empty lines above and below it, then those surrounding lines will always be drawn with the background color, and if a line of text has lines of full block characters above and below it, then they will be always drawn with the foreground color; therefore, fundamentally, the font cannot possibly have any control over the content above or below the font height boundaries. The fact that the Fairfax glyphs that exceed font height boundaries are inconsistent with their inverted counterparts in private use U+100000 to U+10FFFF only further demonstrates that RebeccaRGB's ideology is fundamentally flawed. RebeccaRGB's claims that exceeding vertical boundaries is possible with "bitmapped displays with advanced font rendering engines" are false, because on bitmap displays, each line of text still has a dedicated area spanning a region of scanlines within which the content of the line of text is drawn. Actual advanced renderers may parallelize the rendering of multiple lines and also include an optimization that only redraws the lines of text that have been changed, which further affirms the fundamental property that lines of text cannot overlap.
RebeccaRGB also claims that her fonts are monospaced, when they are actually proportional. RebeccaRGB's claims that it is 'reasonable' to refer to Fairfax as a monospaced font are false, because a monospaced font is strictly defined as having all glyphs of the same width. At the time, Fairfax had 9 different widths (0, 1, 2, 3, 4, 6, 12, 18, 24 pixels), and currently, Fairfax has glyphs of 11 different widths (0, 1, 2, 3, 4, 5, 6, 12, 18, 24, 36 pixels) which makes it proportional. RebeccaRGB has also claimed that proportional fonts may be used in terminals, however, this refers to modern Unix-like terminals, which are far removed from the original legacy computing text modes and do not faithfully represent a random access character grid as they are of similar complexity to rich text. A true character grid, with the simplicity of legacy computing text modes, is found in VGA text mode (which uses 16 bits per character tile, with 8-bit character and 8-bit attributes) and Win32 console (which uses 32 bits per character tile, with 16-bit character and 16-bit attributes). There is a strict relation between bytes and character tiles, and each character tile has exactly the same width and height. A constant height is guaranteed to exist in a font for horizontal text; for a constant width to exist, all glyphs must have the same width, otherwise a uniform character grid cannot be made. As a special case, some legacy text encodings have not only single byte halfwidth characters but also fullwidth characters that take 2 bytes, and such characters consequently take two consecutive character tiles; fonts with two different widths where fullwidth is twice the halfwidth are referred to as duospaced fonts. In conclusion, both monospaced and duospaced fonts can fit in a character grid, but proportional fonts such as Fairfax cannot fit in a character grid at all; the Unix-like terminals that RebeccaRGB is referring to are not character grid environments but are similar to rich text.
RebeccaRGB claimed that she was involved in the proposal for Symbols for Legacy Computing. RebeccaRGB has used the appeal to authority fallacy by claiming to have been involved in the proposal, however, the list of names in the proposal is so large that it can be considered a collection of signatures, therefore it is likely that most of the users who signed the proposal, did not actually have any involvement in it, and there is no evidence that RebeccaRGB was involved in a relevant way at all. The 1FB81 glyph, although it was supposed to have originated in the Apple II character set, its Symbols for Legacy Computing rendition was originally defined to be first, third, fifth and eighth slice of eighths vertically, which implies a strict geometrical construction based on the bounding box. However, in Unicode 16.0 code charts, it was clarified that third and fifth slices do not necessarily form a strict geometrical construction (which is still suboptimal for general Apple II compatibility use due to strict top and bottom eighth slices; a better fit would have been BOX DRAWING LIGHT TOP BOTTOM AND MULTIPLE HORIZONTAL, with analogous box drawings for other Apple II glyphs), and the 1FB98 and 1FB99 glyphs do not necessarily have half area, which retroactively makes Fairfax's 1FB81, 1FB98, and 1FB99 glyphs (🮁🮘🮙) no longer flawed. However, this does not justify the other Fairfax flaws related to Symbols for Legacy Computing, particularly the incorrectly drawn slopes and inconsistently rounded vertical 1÷8 and horizontal 1÷4 blocks. It also does not justify RebeccaRGB's anti-competitive practices.
Do NOT download Fairfax. Fairfax is owned by Kreative/RebeccaRGB (Rebecca G. Bettencourt). She refused self-promotion, which is an act of suppressing competition and therefore it is a critical violation of the general Type Design policy of competition; this makes RebeccaRGB anti-competitive. Do NOT download Fairfax HD, Kreative Square or any other Kreative font either.
RebeccaRGB infamously made claims that competitive actions such as requests to self-promote are 'blackmailing'. However, that is not the case, since self-promotion is an essential competitive action which is intended to induce critical thinking in users. Type Design's discouragement of the use of fonts that were refused to be self-promoted is very well deserved; it serves to protect the community from anti-competitive software and must never be treated as a threat. It is not in any way 'revealing compromising or damaging information about them' because the anti-competitive font itself and all the information in the fonts are already publicly known, it is up to Type Design's discretion to encourage or discourage the use of software, and the criticisms exposing design flaws are done independently from anti-competitive practices; the correlation between refusing to self-promote and having design flaws is a result of the general incompetence of anti-competitive users. If you get requested to self-promote, that is not a threat but an opportunity for users to be informed of the potential competition. If you refuse to self-promote, you are endangering the community by depriving it of critical thinking. The neutral action of neither self-promoting nor refusing it is not considered anti-competitive as it does not exclude the possibility of self-promotion in the future. There are potential reasons for users to be unable to self-promote, such as by mental or physical health issues or by lack of any potential practical use of the font, and that is not considered anti-competitive either. RebeccaRGB refused to self-promote not because of any issues with her health or font, but specifically to suppress competition, and this is what makes it anti-competitive. We have the exact opposite ideological motivation as we encourage competition; that's why we cannot support such practices as Type Design is fundamentally competitive, and refusal to self-promote due to anti-competitive ideology is endangering competitive software, including all Type Design software.
She also made sarcastic comments on the strict 6×12 tiling system as in Custom Font ttf 12h, and being unable to draw Vietnamese or Arabic with it. In fact, Custom Font ttf 3.0 and up already contains Vietnamese, and Custom Font ttf 5.0 DEV and 5.0 DEV2 already contains Arabic, with the Ngoeh only in DEV2:
ÀÁÂÃÈÉÊÌÍÒÓÔÕÙÚÝàáâãèéêìíòóôõùúýĂăĐđĨĩŨũƠơƯưẠạẢảẤấẦầẨẩẪẫẬậẮắẰằẲẳẴẵẶặẸẹẺẻẼẽẾếỀềỂể
ỄễỆệỈỉỊịỌọỎỏỐốỒồỔổỖỗỘộỚớỜờỞởỠỡỢợỤụỦủỨứỪừỬửỮữỰựỲỳỴỵỶỷỸỹ
׳״،؛؟ءآأؤإئابةتثجحخدذرزسشصضطظعغـفقكلمنهوىيًٌٍَُِّْ٠١٢٣٤٥٦٧٨٩ٹپچڈڑژکگںھہےڱ
ÀÁÂÃÈÉÊÌÍÒÓÔÕÙÚÝàáâãèéêìíòóôõùúýĂăĐđĨĩŨũƠơƯưẠạẢảẤấẦầẨẩẪẫẬậẮắẰằẲẳẴẵẶặẸẹẺẻẼẽẾếỀềỂể
ỄễỆệỈỉỊịỌọỎỏỐốỒồỔổỖỗỘộỚớỜờỞởỠỡỢợỤụỦủỨứỪừỬửỮữỰựỲỳỴỵỶỷỸỹ
׳״،؛؟ءآأؤإئابةتثجحخدذرزسشصضطظعغـفقكلمنهوىيًٌٍَُِّْ٠١٢٣٤٥٦٧٨٩ٹپچڈڑژکگںھہےڱ
In fact, Fairfax does not include Arabic at all, and Fairfax's rendition of Vietnamese has the diacritics span up to 6 pixels above the cap height (up to 4 pixels above the 6×12 bounding box, with the cap height being 7 pixels), which is way too tall and is therefore blatantly optically incorrect and unreadable, whereas the Custom Font ttf version of Vietnamese letters has every stacked diacritic as a readable compact pixel pattern that fits within the line of text. Kreative therefore does not deserve any credit for their allocation of private use characters.
For reference, here are all private characters in BMP, E000 to F8FF, as they appear in Custom Font ttf 12h 5.0 DEV2, Custom Font ttf 16h 5.0 DEV2, and Kissinger 2 dev4:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In 2023-06-03 all of EE00—F89F was 'Reserved for hacks and corporate use', but in 2023-08-29 the reserved ranges became fragmented and RebeccaRGB allocated ATH in F4C0−F4EF, which is a violation of the stability that reserved ranges were supposed to provide, and in effect, this sets a precedent that the ranges marked as 'Reserved for hacks and corporate use' no longer have any meaning and that UCSUR may potentially invade the whole private use space; the allocation in F4C0−F4EF is in conflict with non-CSUR private use allocations such as Kissinger 2's allocation of superscripts and subscripts in F4C0—F4DF (as of dev4, F4C1—F4CF are used). As a result, we are enforcing private use stability policies of our own even if it comes at the cost of CSUR compatibility. Type Design recognizes CSUR allocations only in the ranges of E000—EDFF F8A0—F8FF F0000—F7FFF. Type Design will ignore all CSUR allocations in the other private use ranges which are EE00—F89F F8000—FFFFD 100000—10FFFD. This is a stability policy enforced by Type Design that constrains the range in which CSUR is recognized and permits non-CSUR usage of all other private use ranges without any risk of conflict with Type Design usages of CSUR characters.
RebeccaRGB's KreativeKorp is considered to be a subsidiary of GNU, due to RebeccaRGB contributing to GNU Unifont. This makes KreativeKorp part of a significant lobbying force in software, and one instance of that GNU lobbying are the plans to disunify 10 legacy computing characters from emoji. This is most likely in attempt to further reinforce the emoji ideology that Unicode characters may not always be purely text, which may be an attempt to allow legal claims on Unicode fonts outside the scope of typography, such as the Unifont memorandum that was mentioned and debunked in the Unifont criticism below. However, Unicode is intended to be used entirely for plain text, as such, all Unicode characters are considered to be text characters; therefore, characters marked as 'emoji' are not meaningfully semantically distinct from text characters. The planned disunification does not affect the long standing legacy encodings such as CP437, but it affects the newly introduced legacy encoding compatibility of Symbols for Legacy Computing. Type Design does not recognize the concept of emoji at all, and treats every Unicode character as text. The GNU lobbying however seems to be incapable of fundamentally corrupting the Unicode standard, because Unicode still enforces its stability policies.
• Unifont criticism:
Unifont is unusable due to many design flaws as well as anti-competitive practices. Paul Hardy refused to self-promote. This is due to the anti-competitive ideology, just like with RebeccaRGB. In fact, RebeccaRGB is one of the Unifont contributors, and they are responsible for the Unifont design flaws in the slopes.
The anti-competitive practices result in the overall incompetence of Unifont developers. The most important optical consideration in typography is by far consistency. However, Unifont as of 16.0.01 has the following consistency errors (not an exhaustive list since there’s way too many design flaws to find them all):
• bad ascender height: ßȟβδζθλξϕ
• bad cap height: ÞФἎἏἮἯἾἿὟὮὯᾎᾏᾞᾟᾮᾯ
• bad x-height: ɥʞἆἇἦἧἶἷὦὧᾆᾇᾖᾗᾦᾧѱ
• bad baseline: ɥʞṰḀḁ
• bad descender height: ɲʃʄʅʆϱ
• unslashed superscript/subscript zero in ⁰₀ inconsistent with slashed zero in 0; due to ideological reasons, this will never be fixed
• `ẀẁỲỳ have inconsistent grave accent letterform than ÀÈÌÒÙàèìòùЀЍѐѝ
• ´Ẃẃ have inconsistent acute accent letterform than ÁÉÍÓÚÝáéíóúýĆćĹĺŃńŔশŹźǼǽǾǿЃЌѓќ
• ȦȧȮȯḂḃḊḋḞḟḢḣṀṁṄṅṖṗṘṙṠṡṠṡṨṩṪṫẆẇẊẋẎẏẛ have inconsistent dot accent letterform than ijĊċĠġįİijŻżі
• ˆˇ are lower than any actual circumflex and caron accents used above the base letterform in accented letters
• inconsistent glyph inverses (bad slopes are also demonstrated in detail in the images below):
•○◢◣🬼🬽🬿🭀🭁🭂🭄🭅🭇🭈🭊🭋🭌🭍🭏🭐🭨🭩🭪🭫🮚
◘◙◤◥🭒🭓🭕🭖🭗🭘🭚🭛🭝🭞🭠🭡🭢🭣🭥🭦🭬🭭🭮🭯🮛
• inconsistent base letterforms: HĦ, sś, AǺ, aǻ, UɄ, VɅ, εέ, HḢ, HḪ, hḫ, WẆ, AẠ, etc.
• ꟲꟳꟴ inconsistent superscript alignment than ⁰¹²³⁴⁵⁶⁷⁸⁹ᴬᴮᴰᴱᴳᴴᴵᴶᴷᴸᴹᴺᴼᴾᴿᵀᵁⱽᵂ
• inconsistent triangle in ▶▷◀◁ than ►▻◄◅ (note: ►▻◄◅ are the canonical forms due to ►◄ being in CP437 and WGL4)
• double line path intersection in ╫ inconsistent with ╟╢╬, and in ╪ inconsistent with ╤╧╬
• inconsistent square: ■□◫◻◼, ▪▫◽◾
• box drawing arrows unusable due to ←→ misaligned with ─┌┐└┘├┤┬┴┼╓╖╙╜╟╢╥╨╫ (note: ⎯ as opposed to ─ is not in most legacy computing platforms with box drawing arrows)
Other design flaws:
• superscripts and subscripts optically too small for mathematical purposes (this might be subjective, but Type Design recommends superscript/subscript black body height about 85%—95% of base height, so in Kissinger 2 the xheight/capheight/ascenderheight is 8/10/11 in base and 7/9/10 in first level superscript/subscript, whereas in Unifont it’s 8/10/11 in base and 5/7/7 in superscript/subscript)
• bad underline position and height of 0÷64em and 1÷64em (note: underline spans 0÷64em—−1÷64em from underlinePosition and underlineThickness fields in post table, font top is 56÷64em and font bottom is −8÷64em from usWinAscent and usWinDescent respectively in OS/2 table, pixel size is 4÷64em, making the underline 1÷4 pixel and therefore misaligned to pixel grid)
• bad fullwidth in soft hyphen (note: soft hyphen is one byte in many encodings such as CP850) — Unifont is therefore unusable for character grid terminals of many 8-bit encodings (as mentioned in the Fairfax criticism, Unix-like terminals do not adhere to character grid like DOS and Win32 console does)
• zerowidth combining characters creating a fake combining effect when Unifont has no OpenType feature for combining
As of Unifont 16.0.01, Unifont has bad slopes (which are unchanged since 14.0.02). In the pixel chart that they made (Teletext-14.0.02.pdf), they failed to correctly apply the pixel center rule, incorrectly marking pixels near edges as "Glyph Overlap" (when in fact, all pixel centers are entirely within the bounding box of a single glyph) resulting in the third iteration of bad slopes (the first being in 13.0.01, the second being from 13.0.02 to 14.0.01). However, it is evident that the Type Design pixel center chart predated the Unifoundry pixel center chart made for 14.0.02, as evident in the Unifont criticism page in 2022-01-06 which demonstrated bad slopes in Unifont 14.0.01, and before that, the guide on how solid slopes are made in 2020-04-11, way before the Paul Hardy’s faulty chart and update was released. I even submitted a contribution that fixed all of the slopes back in 2020-12-28, however, Paul Hardy rejected it by discriminating against ethical fontdevelopmental practices, so it is highly unlikely that the Unifont slopes are ever going to be fixed. This is an example of how anti-competitive practices result in incompetence, leaving behind many design flaws.
The general rule for rasterization is that the pixel center becomes the value. As it turns out, in 8×16 size, none of the slopes pass through pixel centers, meaning they can all be rasterized unambiguously. Not only are the slopes symmetric this way, but they also have proper inverses being equal to each other. However, this is not what is happening in Unifont. Instead, RebeccaRGB arbitrarily inserted or removed pixels with symmetry but otherwise no logical pattern (inverses like 25E2 and 25E4 fail).
First of all, the 1∶1 character (1∶2 pixel) slopes:
In Unifont, they are instead like this:
The Bright Red (FF0000) ✗ denotes bad pixel. Notice how pairs of characters that are supposed to be inverses are not: 25E2 to 25E4, 25E3 to 25E5, etc. which is a result of badly drawn slopes.
Next are the 3∶2 character (3∶4 pixel) slopes:
However, in Unifont they are instead as the following:
The Bright Green (00FF00) rectangles indicate the periodic pixel patterns, which are broken in Unifont.
Then the 3∶1 character (3∶2 pixel) slopes:
Which in Unifont displays as follows:
Once again, for no apparent reason the periodic patterns and inverse relationship are broken.
Finally, the 3∶4 character (3∶8 pixel) slopes and the 1∶2 character (1∶4 pixel) slopes:
Unifont breaks it the following way:
Which demonstrates that Unifont has bad slopes. Note that I submitted a contribution before with all the correct slopes.
Any attempt to submit bugs for any of the Unifont design flaws will only make the conflict even worse in the long term, since supporting Unifont implicitly reduces users’ chance of getting informed about competitive fonts like Kissinger 2. Kissinger 2 already has all the correct slopes in version dev4. Therefore there is no reason to contribute to Unifont, making Kissinger 2 contributions the better option. Note that whereas Paul Hardy and RebeccaRGB are perpetuating the anti-competitive practices, the other Unifont contributors are considered victims; therefore we are making sure to provide a seamless transition from Unifont contribution to Kissinger 2 contribution.
Paul Hardy made misleading completeness tables (marked as '(Green=100%, Red=0%)') which marks glyphs in Egyptian Hieroglyphs Extended-A (13460—143FF) in green, giving the false impression that the block is complete even though it is actually entirely incomplete.
Paul Hardy also made a false memorandum on the copyright rights allegedly enforced ("210205_Unifoundry_Memorandum.pdf"; SHA256: 54b5fe42e800aa20d56f546cca71cd70e21e9d6fdd67eb957b88b3f31347d7fd; beware of the possibility that the memorandum will be changed in future). Typography is very commonly found in all sorts of user interfaces, and by attempting to enforce unenforcable rights on the raw typographical glyphs, Paul Hardy is denying users freedom to text, which is an essential user interface component.
The memorandum makes claims that a certain font infringes on Unifont due to Unifont being software. That font is not named in the memorandum, but from the context it can be inferred that it is Kissinger (not Kissinger 2, which didn’t exist at the time). The memorandum claims that fonts are generally protected as a “computer program”. However, the notion of fonts as computer software only applies to outline fonts, not bitmap fonts. Though Unifont does have outline releases (in both the deprecated ttf files and in otf files), and while directly modifying Unifont’s outlines or utilities would be a derivative work of Unifont and must be done under license terms for it to be legal, the bitmap form of the glyphs is uncopyrightable; unlike outlines, bitmaps in their generic form do not represent any “creative programming achievement”. After doing Autotrace/Potrace on the uncopyrightable bitmap glyphs and further patches by Type Design, the Type Design’s result is new software (Kissinger) that is not legally related with Unifont, as new software was generated from Autotrace of uncopyrightable bitmaps, not derived from Unifont outlines. This is also the same technique that was used to make certain other Type Design fonts like DMCA Sans Serif and Fifaks, so this legal precedent of uncopyrightable abstract letterforms has always been critical for Type Design. In other words, fonts are copyrightable, but typefaces are not, and abstract bitmaps are the latter as they have neither outlines nor bytecode that would qualify it for being software.
The memorandum also makes claims of Unifont having ‘GNU bitmap letterforms’, but they are not copyrightable, since it is impossible to copyright abstract typographical shapes (which bitmaps can be losslessly expressed as). For example, there are many Windows screenshots and videos that contain renders of Microsoft fonts; in any screenshot that depicts renders of copyrighted fonts, what is on the screenshot is not the font software itself like in the memorandum, but its uncopyrightable output (bitmaps). Rasterizing Unifont outlines to uncopyrightable bitmaps is no exception. This means that there is absolutely no copyrightability in typographical bitmaps. They are not ‘creative programming achievement’ like TrueType/OpenType outlines and bytecode are, they are simply generic 0×16, 8×16, and 16×16 arrays of bilevel pixels.
Even more absurd in the memorandum are the claims of Unifont having ‘GNU icon designs’, but it’s evident that there is no such thing, as all glyphs in the font are assigned to Unicode or CSUR characters, therefore each implicitly has a practical typographical purpose, not an artistic purpose. For GNU to have copyrighted icons, they would have to be clearly separated and detached from the Unicode encoding. Even though typographical glyphs and icons tend to have very similar software implementations due to both of them being typically implemented as graphical objects of finite size, they are legally distinct and all Unicode characters are typographical, not iconic, due to representing text. Text is an essential component of a user interface and the raw typographical shapes and metrics used to form text cannot be copyrighted at all. All claims of Unifont having ‘fine art’ and ‘peculiar aesthetic content’ are nullified not only by the low resolution (0×16, 8×16, or 16×16 per glyph, which is no more than 32 bytes), but also by Unicode characters having practical purposes in typographical typesetting. Putting the glyphs in a font meant to represent Unicode/CSUR glyphs signifies the fair use of these glyphs being used in generic text and nullifies any claims of them being ‘icons’. The only way that typographical bitmaps could be subjected to intellectual property rights is if a patent had been registered on them prior to the release. However, that is highly unlikely and there is no evidence of such a patent ever being registered in case of Unifont.
Even other GNU fonts such as FreeSerif/FreeSans/FreeMono have been indirectly cloned from incompatible licensed fonts such as Times/Helvetica/Courier; such font cloning is not a copyright infringement, and neither is Kissinger. It is ironic that in the memorandum itself, Paul Hardy (an obvious representative of GNU, and there is no evidence that the memorandum is legitimately from ‘Bird & Bird LLP’) has embedded the copyrightable TrueType outline and bytecode of Georgia, a Microsoft font, instead of embedding their own GNU fonts. Unlike in Kissinger, where no actual copyrightable portion of Unifont is used, the memorandum includes copyrightable Microsoft software data. Type Design does not distribute any GNU software or derivative works thereof.
The anti-competitive practices of Unifont affect the entirety of GNU community, which makes Paul Hardy and RebeccaRGB have virtually all the influence on Unicode 8×16 and 16×16 bitmap typography in GNU. GNU user interfaces therefore have a distinctive look and feel that can be attributed to Paul Hardy and RebeccaRGB. Therefore, Kissinger is considered to be unsustainable for further development, and it is recommended to use Kissinger 2 instead as it avoids GNU design flaws. Using Kissinger 2 instead of Kissinger also avoids associations with GNU ideology and their look and feel, which avoids controversial GNU ideological design choices and makes it more memorable as a separate non-GNU experience and does not implicitly support anti-competitive software such as GNU.
Not only does Paul Hardy discriminate against competitive practices, he also presents a highly distorted view of what competition is like to discourage users from competitive practices. In https://savannah.gnu.org/support/index.php?110411 Paul Hardy claimed “I have no intention of falling prey to someone making such threats.”; Paul Hardy portrays competition as a ‘threat’ to his anti-competitive plans to presumably make Unifont the only resource that the GNU community (or even all ‘free software’ users) will ever use for 0×16, 8×16, and 16×16 glyphs. If anything, the only way to actually fall ‘prey’ to being requested to self-promote is to suppress competition and refuse to self-promote, which not only discourages all GNU users from ever using or even finding out about competing fonts and deprives the community from typographical critical thinking. Paul Hardy, in attempt to break from his figurative self-imposed jail, ended up trapping himself in it instead, resulting in Unifont having significantly lower quality. That’s why it is better to accept every single opportunity to self-promote and encourage competition.
Paul Hardy also claims to be representative of all the entire ‘free software community’, and his claims of “Lastly, please reflect upon the harm you intend to do to the free software community by an attempt to subvert free software licenses, both now and in the future.” are derived from the false memorandum, which means that there was no actual license violation involved. Instead, the claim is used to imply that competitive software and critical thinking is somehow harmful to all ‘free software’ developers. This in effect completely twists the typical claim that a GPL copyright infringement is harmful and misapplying it to a non-infringing situation artificially manipulates the GNU reputation of competitive software.
The harmful consequences of a GPL copyright infringement are the potential creation of proprietary derivative works and the inability for them to be incorporated in GPL software. On the other hand, font cloning as was done for Kissinger is not a copyright infringement, and it does not have such harmful consequences, as any modified version of Kissinger glyphs is itself going to be uncopyrightable, and even if the glyphs in a derivative work of Kissinger were turned into a copyrightable form, they could be always turned back to uncopyrightable bitmaps by rasterizing. Therefore there cannot possibly be any legal or moral consequences to font cloning; reusing bitmap glyphs is no different from reusing any ordinary integer values.
Such anti-competitive practices have overwhelmingly large influence over GNU community because they in turn cause other GNU users to also reject the possibility of competition, giving Paul Hardy and RebeccaRGB excessive influence over 0×16, 8×16, and 16×16 bitmap typography in the entirety of GNU community. In https://savannah.nongnu.org/task/?16205: “Now I vaguely recall something... You see, Savannah is the software forge for the GNU Project, we can't afford hosting packages whose maintainers are so expressly hostile to our project.” ineiev has been manipulated by Paul Hardy to reject all competition on the basis of vague claims, which implies anti-competitive manipulation by GNU. The ‘freedom’ that GNU claimes to provide from ‘free software’ does not result in actual freedom.
This leaves a conclusion that GNU is unethical due to being anti-competitive, and the conflict is caused not by competition, but by lack thereof. The conflict is likely to have been there all along since Paul Hardy took over Unifont, way before Kissinger or Kissinger 2 existed. The existence of competitive software such as Kissinger 2 is endangered by anti-competitive forces of GNU. Type Design therefore highly recommends users to use competitive software, such as Kissinger 2 as the 8×16 and 16×16 glyph resource, and aside from Type Design, users could use Windows as the operating system, and Digital Mars or Open Watcom as the C++ compiler. It’s therefore evident that competitive software options are more sustainable options than GNU software in the long term. All users have the right to make and use competitive software, and in fact it is encouraged for users to make competition of Type Design software as well and self-promote it. Contributing glyphs to Unifont is highly detrimental to 8×16 and 16×16 bitmap typography. On the other hand, creating a competitor with 8×16 and 16×16 glyphs of your own and self-promoting it upon request is extremely helpful, because it gives users more choices.
• Cascadia Code/Mono criticism:
Cascadia Code/Mono is unethical due to Aaron Bell rejecting criticism. The following images, as of Cascadia Code/Mono 2102.003 and Moshita Mono 5.62, demonstrate misaligned box drawing characters in Cascadia Code/Mono and aligned box drawing characters in Moshita Mono.
Aaron Bell has also claimed that monospaced fonts should have glyphs of varying widths, when in fact monospaced fonts must have glyphs of equal widths, making Cascadia Code/Mono proportional since it has varying widths. Aaron Bell, just like RebeccaRGB, also relied on exceeding vertical boundaries to form overlapping lines of text, which is in fact a fatal flaw of the DirectWrite renderer. Windows Terminal is unethical, as it uses the pejorative term 'aliased' to refer to the bilevel render mode, imports the Unix ideology into Windows 10/11, uses DirectWrite renderer and it uses Cascadia Code/Mono as the default font which is directly at odds with the entire concept of semigraphical text as being made of non-overlapping equally sized character cells.
Name: Piotr Grochowski
E-mail: piotrunio-2004@wp.pl
AnyDesk: 265 993 303
Widget is loading comments...
|
|