«

Hi all,

I have a strange problem using the following setup :

-Red Hat Linux 9 (using latest Fedora kernel)
-Java 1.4.1_02
-IntelliJ IDEA 3.0.4

I use IDEA's option to antialias the font in the editor and
it looks really very good on my screen (Dell 20" 1600x1200) :
can't code anymore without a nice flat screen, without
a digital cable between the gfx card and the monitor, and
without clean AA

Anyway, don't want to start a flame war between the pro-AA
and the anti-AA, here's the problem.

Sometimes, maybe once every five times I lauch IDE, the AA
fonts don't look good, while the other times they are perfect
(well, I find them perfect on my screen at least).

Here are two small screenshots showing the problem :

http://users.skynet.be/fa268924/aliasing.html

Shots were taken on a screen using "RVB" pixels setup, though
I think it doesn't matters has AA here looks like "simple" AA
and not "subpixel"/"rgb decimation" rendering (btw does
Java 1.4 AA use subpixel rendering at all under Linux !?).

You can see the problem on the text "response.encodeURL".

Once I launch IDEA and the fonts are good, they stay good
for the whole session... If I close IDEA and reopen it,
usually it's OK but sometimes they're not looking good.

Typically, if I open the editor and notice that the fonts
are ugly I just close the editor and relaunch it, so it's
really not a big deal, but I wonder if anybody has had the
same kind of problem ?

I decided to post it after reading the recent thread about
aa fonts in Java 1.4 because the "ugly" version looks a
bit like the "ugly" version I sometimes get...

Don't know if it's related to Java 1.4, to IDEA or to
Linux so I apologize if it's OT,

Thanks in advance for any thoughts on this,

Jean

Jean Lutrin wrote:

> Hi all,
>
> I have a strange problem using the following setup :
>
> -Red Hat Linux 9 (using latest Fedora kernel)
> -Java 1.4.1_02
> -IntelliJ IDEA 3.0.4
>
> I use IDEA's option to antialias the font in the editor and
> it looks really very good on my screen (Dell 20" 1600x1200) :
> can't code anymore without a nice flat screen, without
> a digital cable between the gfx card and the monitor, and
> without clean AA
>
> Anyway, don't want to start a flame war between the pro-AA
> and the anti-AA, here's the problem.
>
> Sometimes, maybe once every five times I lauch IDE, the AA
> fonts don't look good, while the other times they are perfect
> (well, I find them perfect on my screen at least).
>
> Here are two small screenshots showing the problem :
>
> http://users.skynet.be/fa268924/aliasing.html
>

<snip>

On my 19" CRT at 1280x1024 those screenshots look exactly the same...

Matt

--
Not so interesting...
http://www.mpcontracting.co.uk

"Matt Parker" <> wrote in message
news:402bd997$0$14754$.. .
> Jean Lutrin wrote:
>
> > Hi all,
> >
> > I have a strange problem using the following setup :
> >
> > -Red Hat Linux 9 (using latest Fedora kernel)
> > -Java 1.4.1_02
> > -IntelliJ IDEA 3.0.4
> >
> > I use IDEA's option to antialias the font in the editor and
> > it looks really very good on my screen (Dell 20" 1600x1200) :
> > can't code anymore without a nice flat screen, without
> > a digital cable between the gfx card and the monitor, and
> > without clean AA
> >
> > Anyway, don't want to start a flame war between the pro-AA
> > and the anti-AA, here's the problem.
> >
> > Sometimes, maybe once every five times I lauch IDE, the AA
> > fonts don't look good, while the other times they are perfect
> > (well, I find them perfect on my screen at least).
> >
> > Here are two small screenshots showing the problem :
> >
> > http://users.skynet.be/fa268924/aliasing.html
> >
>
> <snip>
>
> On my 19" CRT at 1280x1024 those screenshots look exactly the same...
>
> Matt

And mine which is the same.

Jean Lutrin wrote:

> Anyway, don't want to start a flame war between the pro-AA
> and the anti-AA,

Aww...


> Sometimes, maybe once every five times I lauch IDE, the AA
> fonts don't look good, while the other times they are perfect
> (well, I find them perfect on my screen at least).
>
> Here are two small screenshots showing the problem :
>
> http://users.skynet.be/fa268924/aliasing.html

I haven't seen any helpful replies to this yet. This one won't be very helpful
either

Looking at the images, I'd say that the AA engine has been given slightly
different typefaces to work on. Some of the characters come out bit-for-bit
identical, which suggests that it's not a problem with the AA engine per se --
it's there and it's running normally. Many of the differences seem to center
around the cross-bars like the horizontal of a lower -case 'h' or 'e'. It
looks as if the engine has been told either that the crossbars are thinner in
your "wrong" picture, or that they are slightly differently positioned.
Whichever, it doesn't use as much black to draw them. The consistancy suggests
that the AA engine is reflecting a consistant change in the definition of the
typeface itself (the position and thickness of the crossbars is one of the
things that tends to be consistant across the glyphs in a typeface, naturally).

IIRC, X uses a system where you ask for a a font using a pattern spec, and it
tries to find the closest match to the spec. OTOH I don't know if Java/IDEA
uses X fonts at all, or whether it's entirely done in Java graphics. In either
case, though, it seems that the IDE is sometimes picking up a different
typeface, or is rendering the same typeface with slightly different settings
(fractional font-metrics on/off for example) for some reason.

As I said, not very helpful. But at least I can confirm that you are not
seeing things...

-- chris

> As I said, not very helpful. But at least I can confirm that you are not
> seeing things...

Thanks for the infos, your explanations are interesting.

Thanks also for confirming that I am not "seeing things". I knew it, but
still: having two people saying, after my description of the problem,
that the two screenshots looks the same is frightening to say the least.

I know I do have a very good eyesight and a good screen, but if those
two screenshots are looking exactly the same on several people's screens,
methinks there's something wrong :-/

I used to work on high-end Sony Trinitrons for a few years (changing quite
often, as they tend to get used pretty fast) and now I've switched to a
flat screen (with a digital cable, this is important) and I think that
for coding there's simply no match. Less headaches for me too (actually,
no more headaches due to the monitor).

Thanks for everybody's feedback,

Jean

Jean Lutrin wrote:
>>As I said, not very helpful. But at least I can confirm that you are not
>>seeing things...
>
>
> Thanks for the infos, your explanations are interesting.
>
> Thanks also for confirming that I am not "seeing things". I knew it, but
> still: having two people saying, after my description of the problem,
> that the two screenshots looks the same is frightening to say the least.

Well, parts of it were the same, and part's weren't. That made it trickier.


>
> I know I do have a very good eyesight and a good screen, but if those
> two screenshots are looking exactly the same on several people's screens,
> methinks there's something wrong :-/
>

Fuzzy CRT's and people didn't use Pixie, XMag, Zoomit, etc.



> I used to work on high-end Sony Trinitrons for a few years (changing quite
> often, as they tend to get used pretty fast) and now I've switched to a
> flat screen (with a digital cable, this is important) and I think that
> for coding there's simply no match. Less headaches for me too (actually,
> no more headaches due to the monitor).

I'd have to agree with that.

And be sure to look at all the numbers. Brightness, contrast, horizontal
and vertical viewing angles...

Jean Lutrin wrote:

> I used to work on high-end Sony Trinitrons for a few years (changing quite
> often, as they tend to get used pretty fast) and now I've switched to a
> flat screen (with a digital cable, this is important) and I think that
> for coding there's simply no match.

Agreed entirely. CRTs are fine for watching movies, but not for actual work.

I'm hoping that I'll never have to stare at code on a CRT again.

-- chris