[9fans] ttf2subf (again)

[9fans] ttf2subf (again)

[Joel Salomon]

Charles Forsyth wrote:

>>yes, we are aware of the fact that Charles Forsyth has done ttf2subf long
>>time ago, but that code never made it out and we have no idea whether it
>>did antialiased fonts (reasonable assumption is that it did)...
>i gave it to several people; i don't know what happened to it then.

If you mean you lost the code, I still have the tarball you sent me
(including the freetype port) saved somewhere. I tried it on the
Code2000/Code2001 fonts (trying to extract the elvish glyphs :-) but never
got it to work. Anyway, I have the freetype port by Forsyth and can mail
it out if anyone needs it.

--Joel

Re: [9fans] ttf2subf (again)

[YAMANASHI Takeshi]

I've done something about the ttf font server.

The idea is to generate a requested font
(ex. /lib/font/bit/ttf/luxirr.20.font) automatically
using coresponding ttf font (ex. /lib/font/ttf/luxirr.ttf).

Source files
http://p9c.cc.titech.ac.jp/plan9/fontfs.tgz

Comments
http://p9c.cc.titech.ac.jp/plan9/fontfs.html
--
YAMANAHSHI Takeshi

Re: [9fans] ttf2subf (again)

[matt]

>
>
>specs), but then again it might be nice to be able to specify "12pt
>bold" without worrying about the exact typeface...
>

when I read about fonts I envisaged the server somehow accepting CSS
definitions

Re: [9fans] ttf2subf (again)

[rog]

> I kind of feel implement it as a file server
> sits on somewhere like /lib/font/bit/ttf.  When a
> certain font like /lib/font/bit/ttf/R.7.1 is requested
> to be open(2)ed and doesn't exist, the server creates
> it using ttf2subf and returns the resulted bit map font.

the question is how much information to embed in the path name...  and
what does one do about partial matching.  short names are desirable
(we don't want to end up with something that looks like X11 font
specs), but then again it might be nice to be able to specify "12pt
bold" without worrying about the exact typeface...

Re: [9fans] ttf2subf (again)

[Martin C.Atkins]

Hi Chris,

On Tue, 12 Aug 2003 19:54:27 +0100 "Chris Hollis-Locke" <chris@hollis-locke.com> wrote:
>...
>
> My aim is to produce a scalable font server.

Do you mean a "font server that is scalable", or a "server
for scalable fonts"?

Either could potentially be interesting, depending on one's
perspective... :-)

Martin
--
Martin C. Atkins			   martin@mca-ltd.com
Mission Critical Applications Ltd, U.K.	   http://www.mca-ltd.com{/,/martin}

Re: [9fans] ttf2subf (again)

[YAMANASHI Takeshi]

> My aim is to produce a scalable font server.

I'm interested in the implementation of the server.

I kind of feel implement it as a file server
sits on somewhere like /lib/font/bit/ttf.  When a
certain font like /lib/font/bit/ttf/R.7.1 is requested
to be open(2)ed and doesn't exist, the server creates
it using ttf2subf and returns the resulted bit map font.

The beauty of Plan 9 file server.
--
YAMANASHI Takeshi

Re: [9fans] ttf2subf (again)

[Latchesar Ionkov]

On Tue, Aug 12, 2003 at 07:54:27PM +0100, Chris Hollis-Locke said:
> Lucho wrote...
> >...
> >currently there is no support for these fonts in Plan9. The images are
> >converted to GREY8 before they are used as a mask.
>
> Yes - and a GREY8 mask acts as the alpha channel.
>
> Other things...
>
> sub-pixel rendering is used to apportion the amount of alpha set in the
> lower-res pixels.  In the case of freetype, this is the grey level (how black the pixel is)
> this maps directly to the use of a GREY8 mask as alpha.

I guess I didn't use "subpixel rendering" term correctly. By subpixel
rendering I mean the rendering for LCD displays that takes into account the
fact that a pixel on the LCD screen consists of three colored squares (RGB
or BGR). The result of such rendering is an image with red, green and blue
channels that should be used separately as alpha channels for the red, green
and blue colors when drawing the character. Theoretically this could achieve
three times better horizontal resolution on LCD displays. Practically it is
less (otherwise the letters look differently colord), but it is still much
better than the normal antialiased rendering.

> I have ported freetype to Inferno, as a system module, and have limbo tools for generating
> font and subfont files.
>
> My aim is to produce a scalable font server.

I thought about that too, it would be great if somebody else does it first :)

Thanks,
	Lucho

Re: [9fans] ttf2subf (again)

[Chris Hollis-Locke]

Lucho wrote...
>...
>currently there is no support for these fonts in Plan9. The images are
>converted to GREY8 before they are used as a mask.

Yes - and a GREY8 mask acts as the alpha channel.

Other things...

sub-pixel rendering is used to apportion the amount of alpha set in the
lower-res pixels.  In the case of freetype, this is the grey level (how black the pixel is)
this maps directly to the use of a GREY8 mask as alpha.

I have ported freetype to Inferno, as a system module, and have limbo tools for generating
font and subfont files.

My aim is to produce a scalable font server.

Chris.

Re: [9fans] ttf2subf (again)

[Latchesar Ionkov]

Just couple of comments ...

Although ttf2subf supprots subpixel rendering and generates RGB24 images,
currently there is no support for these fonts in Plan9. The images are
converted to GREY8 before they are used as a mask. Also, only RGB rendering
was really tested. I am not sure the VRGB and VBGR support is correct.

As Charles Forsyth already mentioned antialiased fonts are not something new
in Plan9 -- there are already alot of them (althought they have only 4
levels of grey). The kudos should to go to the authors of the *draw libraries
that made the antialiased fonts support so flawless. If only they thought
about the subpixel stuff too ... ;)

Thanks,
	Lucho

On Tue, Aug 12, 2003 at 10:13:50AM -0600, andrey mirtchovski said:
> Luchezar Ionkov has ported freetype to plan9 and has a working ttf2subf
> conversion utility. The end result is that we now have antialiased fonts in
> Plan 9.
>
> download the utility and some pre-converted fonts (under the XFree86
> license) here:
>
> 	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/freetype/
>
> also available are screenshots at the bottom of the page. i've put a gif of
> one of the screenshots, in case you're unable to view the png's under
> Plan 9 (png(1) is buggy):
>
> 	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/freetype/cyber.gif
>
>
> yes, we are aware of the fact that Charles Forsyth has done ttf2subf long
> time ago, but that code never made it out and we have no idea whether it
> did antialiased fonts (reasonable assumption is that it did)...
>
> andrey

Re: [9fans] ttf2subf (again)

[Charles Forsyth]

>>yes, we are aware of the fact that Charles Forsyth has done ttf2subf long
>>time ago, but that code never made it out and we have no idea whether it
>>did antialiased fonts (reasonable assumption is that it did)...

i gave it to several people; i don't know what happened to it then.

also i'm fairly sure some existing Plan 9 fonts or subfonts are antialiased.

[9fans] ttf2subf (again)

[andrey mirtchovski]

Luchezar Ionkov has ported freetype to plan9 and has a working ttf2subf
conversion utility. The end result is that we now have antialiased fonts in
Plan 9.

download the utility and some pre-converted fonts (under the XFree86
license) here:

	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/freetype/

also available are screenshots at the bottom of the page. i've put a gif of
one of the screenshots, in case you're unable to view the png's under
Plan 9 (png(1) is buggy):

	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/freetype/cyber.gif


yes, we are aware of the fact that Charles Forsyth has done ttf2subf long
time ago, but that code never made it out and we have no idea whether it
did antialiased fonts (reasonable assumption is that it did)...

andrey