[links-list] Re: font problem and fix

aludal aludal at softhome.net
Sun Nov 10 15:19:27 PST 2002

В сообщении от 10 Ноябрь 2002 14:05 вы написали:
> > good hinting capabilities over a wide range of screen resolutions
> > (unlike, say, M$ Times New Roman, Monotype, or Adobe Times, etc)
> I do not support any hinting as it would require ambedding a font
> interpreter into the browser. And font interpreter is too complicated
> thing to ever work reliably. Second, making a new PNG is easy (you just
> take GIMP). I don't know how to make new glyph in TT and existing fonts
> are usually copyright-burdened so they are useless.
you don't need to support hinting, all I was saying is that conversion of a 
(TT, Type1, and the like vector/curves) outline could be made with good pixel 
grid fit (i.e., good hinting) or with a bad one.
On "easyness" of making a new PNG: are you telling me you drew all the 
outlines of Century School BT (by the way, ITC, or Agfa/Lynotype copyrighted, 
can't tell exactly which foundry as ~100by55 PNG files lost that fine 
distinctions) right out of your head, on the blank field? I somehow doubt it.
And yes, using GIMP for this is easy, I agree. When you want to produce one, 
or two, or even 200 PNGs, one by one.
PfaEdit is perfectly capable of producing a new glyph in TT (if "a new glyph" 
is ever needed, that is; so far, Titus is complete enough. PfaEdit is also 
capable of producing PNGs, one by one, out of TT font. As long as in your 
work you haven't access to, and didn't use so called Master outlines of a 
given font, but merely used their TT renditions, then retouched your 
resulting PNGs  manually, it's not copyrighted font anymore. It's your font 
now (BTW, you'd better remove that dir names like "century_school*" from the 
Links releases unless you paid to ITC around $50K for use of their 
copyrighted stuff in unlimited distributions.

> > of code, could it amount to the 'reverse-engineering' needed? I can't
> > imagine it could be very hard to code it as even UI option, but you guys
> > having better things to do like bugbusting, etc. But believe me, it's not
> > only a question of ugliness of dark grey background. The 'hardwired'
> > antialiasing of Links glyphs was designed against white background, so a
> > big part of that
> I have been designing the antialiasing myself and assure you, it isn't
> "designed against white background". It generates a resampled alpha mask
> which is then used to combine foreground and background color.
OK, so what you're saying essentially is that your AA is made on-the-fly and 
is background/letter body-color adjustable, right?

> > smoothing grey 'subpixels' mingles with the darkgrey background
> > effectively destroying font outlines in the result. You guys are talking
> > about embedded Links for 320x200 screens of PDA, mobiles, etc. This
> > darkgrey background problem would be much worse for those silly toys.
> Wouldn't. You apparently don't understand what antialiasing is.
Look, when redesigning/retouching RusCyr and designing RusCyrBolds, I manually 
added hundreds of 'pseudoAA' grey pixels to pngs, or stretched/squeezeed 
these pngs to produce better smoothing ones, or delete stray ones. Without 
this retouch and without those grey pixels, I assure you, no 'resampled alpha 
mask' can correct an outline -- they looked ugly.
> > So, my suggestions are:
> > 1) it needs to be user-configurable;
> > 2) for the web pages of 'white/light letters on dark/black
> > background'-sort, there should be user selectable option to inverse these
> > to 'black letters on white background'. This request stems from the fact
> > that those AA subpixels are not symmetrical, either, so readabilty
> > suffers even more.
> All pixels are symmetrical on the screen. I don't know what are you
> talking about.
It's an optical illusion: you inverse the colors of your foreground, 
background, and grey pixels, and you see your visible outlines getting bolder 
or thinner. Then, in lower resolutions and/or lower bpp counts this kind of 
inversion can even destroy the outlines. Or restore them to a reasonable 
Unsubscribe: send email to links-list-request at linuxfromscratch.org
and put unsubscribe in the subject header of the message

More information about the links-list mailing list