Version: (using KDE KDE 3.2.1)
Installed from: Debian testing/unstable Packages
Note: this is a forward of the following Debian bug report:
The original report follows:
This will seem unbelievably petty, but once you notice it it may bug
The drop shadow on fonts used on the desktop is incomplete. It seems
that text is rendered in 8 positions around the original position to
create a broad outline, but not in the center of the original
position. The result is that if you use a font that has a 1-pixel
period, it makes a hollow box for the shadow instead of a solid
box. There is no way that shadow would match the period.
Created attachment 5251 [details]
I think this patch is the proper fix.
---------- Forwarded Message ----------
Subject: Re: Fwd: [Bug 77841] kdesktop: drop shadow on desktop icon text is incomplete
Date: Wednesday 17 March 2004 19:29
From: Laur Ivan <firstname.lastname@example.org>
To: David Faure <email@example.com>
On Wednesday 17 March 2004 17:13, you wrote:
> Can you comment on this patch?
I had the specific pixel omitted as there are cases when it creates rather
visible alasing. The case I tried to avoid is of anti-aliased font with all
qGray(pixel) == 255 and a 3-pixel wide shadow. If you get the shadow only,
the difference between pixels belonging to the character and adjacent pixels
is quite high when you include 3*qGray(pixel) as the adjacent pixels will
constantly miss that. This is reflected quite well in a character's diagonal
strokes when the shadow is displaced 1+1pixels. The original character would
and with shadow:
where S are the pixels associated to the character's pixels and "." are the
faded pixels adjacent to S.
IMO it's more a matter of taste than a bug :). Also, the patch involves
changing the multiplier setting.. which is quite unpleasant for people who
have customised their own shadow in the days of 3.1.x and are using that..
However, the patch makes the algorithm 100% accurate. If you decide to put it
in, the 3.0 factor should be changed to something according to a gaussian
distribution in relation to the 1.0 and 2.0 factors.
David Faure writes:
> However, the patch makes the algorithm 100% accurate. If you decide
> to put it in, the 3.0 factor should be changed to something
> according to a gaussian distribution in relation to the 1.0 and 2.0
I am going to leave this decision at your discretion. I haven't read
about or worked with graphics stuff enough to be able to do it.
Please let me know what you decide, though, so I know what to do with
the debian bug about this.
Kdesktop is no more mantained. Fortunatly this bug seems not to be valid for KDE 4. Please reopen if this bug is not a kdesktop one (and it is not solved) or it can be reproduced on KDE 4.
Bug closed. Kdesktop is no more mantained.