[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    [Fwd: Re: What do you consider more important? (graphics)]
From:       mosfet <mosfet () mandrakesoft ! com>
Date:       2000-02-26 0:16:40
[Download RAW message or body]

This was an earlier mail I was sent.

-------- Original Message --------
Subject: Re: What do you consider more important? (graphics)
Date: Fri, 25 Feb 2000 16:04:52 -0800
From: Keith Packard <keithp@suse.com>
To: mosfet <mosfet@mandrakesoft.com>
CC: kde-core-devel@kde.org, keithp@suse.com


> I couldn't resist to ask about that AA stuff and he said that the AA stuff
> is *not* dead. Keith Packard <keithp@suse.com> *is* working on it. He said
> that it will be in post XF40 though.

I am working on AA graphics, and yes text AA is the principle reason. 
The 
API isn't as easy as I first thought.  AA graphics are typically 
approximated with alpha blending, but which style of alpha blending 
depends on the background in use and the order primitives are rendered
in, 
so we end up needing to let the application choose from several 
alternatives.

There are other ways to approximate AA graphics, one choice that SGI 
currently implements in hardware is a super-sampled frame buffer.  We 
could do the same in software but would lose hardware acceleration.

Additionally, we'll want to add sub-pixel coordinates for finer control
of 
glyph positioning, and also new rendering primitives for converting 
characters into graphical objects.

My plan is to build a set of interoperating systems so we can work on
each 
part separately.  First comes alpha blending, next comes new rendering 
primitives and finally comes AA graphics.  The idea is to hook app
writers 
with the speed of accelerated alpha blending (~100 times faster than 
software) and get them excited enough to help drive the rest of the 
standard.  I can design a rendering system in isolation, but I suspect 
we'll like the results better if we work together.

This work is moving slowly at this point; XFree86 is busy getting 4.0
done.

I'll be giving a short presentation at Usenix in June about this work
and
hope to have something to show people at that time.

-keith

[prev in list] [next in list] [prev in thread] [next in thread]