Re: Some xterm-256color face colors too bright?

Re: Some xterm-256color face colors too bright?

[Eli Zaretskii]

> From: stktrc <stktrc@yahoo.com>
> Date: Mon, 31 Jan 2005 15:07:59 +0200
> 
> The default colors of some faces become very bright, bordering on
> too bright to be readable, when using an xterm-256color terminal
> with light background (correctly detect in frame-background-mode).
> Such faces are for example font-lock-string-face and
> compilation-warning-face.

If you type "M-x customize-face RET font-lock-string-face RET", what
color name do you see in the Custom buffer?

The default color for that face on light-background displays with more
than 88 colors is "RosyBrown".  I cannot see how this could be so
bright as to annoy you.  Could it be that the background color is
actually the problem (Emacs doesn't touch the background color in a
non-windowed session, it uses the color that it found)?

You might also start a windowed Emacs session with a light background
and compare the colors for that face in the two sessions, both for
background and foreground colors.  If you do that, what do you see?

> I guess that this is not related to the individual face selections,
> but rather to the algorithm by which colors are mapped onto XTerms
> colors.  Any hints on which parts of Emacs to look in to tune this?

The code that does it is in lisp/term/xterm.el.  But I don't think
that code is to blame: all it does is register the 256 colors with
their RGB values, which cannot possibly cause such problems.

Re: Some xterm-256color face colors too bright?

[Luc Teirlinck]

Eli Zaretskii wrote:

   I cannot see how this could be so bright as to annoy you.

How bright a given color looks to a given person can vary _widely_
from person to person.  It of course also depends on the monitor, as
well a a variety of other factors.

Sincerely,

Luc.

Re: Some xterm-256color face colors too bright?

[Eli Zaretskii]

> Date: Mon, 31 Jan 2005 14:22:39 -0600 (CST)
> From: Luc Teirlinck <teirllm@dms.auburn.edu>
> CC: stktrc@yahoo.com, emacs-devel@gnu.org
> 
> Eli Zaretskii wrote:
> 
>    I cannot see how this could be so bright as to annoy you.
> 
> How bright a given color looks to a given person can vary _widely_
> from person to person.

If this is personal, then the colors in the windowed session of Emacs
will look annoyingly bright as well (to the same person).  Which was
one of the points in my message.

Re: Some xterm-256color face colors too bright?

[Eli Zaretskii]

> From: stktrc <stktrc@yahoo.com>
> Date: Tue, 01 Feb 2005 13:48:49 +0200
> 
> Comparing terminal RosyBrown and windowed RosyBrown as background
> colors gives that terminal RosyBrown is slightly brighter, but the
> difference is not huge.
> [...]
> I want to stress that the brightness of the colors in the terminal are
> not just an aesthetical annoyance or anything like that: it borders on
> making the text unreadable.  In windowed Emacs, everything is fine.

??? These two paragraphs seem to contradict one another: the first one
says that the difference between the windowed and non-windowed session
is small (which is expected, as the former one has more colors than
256), while the latter says that the difference is large (``borders on
unreadable'' vs ``everything is fine'').  I wonder how can this
happen.  Can you shed some light on this?

Re: Some xterm-256color face colors too bright?

[Stefan Monnier]

>>>>> "Eli" == Eli Zaretskii <eliz@gnu.org> writes:

>> From: stktrc <stktrc@yahoo.com>
>> Date: Tue, 01 Feb 2005 13:48:49 +0200
>> 
>> Comparing terminal RosyBrown and windowed RosyBrown as background
>> colors gives that terminal RosyBrown is slightly brighter, but the
>> difference is not huge.
>> [...]
>> I want to stress that the brightness of the colors in the terminal are
>> not just an aesthetical annoyance or anything like that: it borders on
>> making the text unreadable.  In windowed Emacs, everything is fine.

> ??? These two paragraphs seem to contradict one another: the first one
[...]
> unreadable'' vs ``everything is fine'').  I wonder how can this
> happen.  Can you shed some light on this?

He also said:

   Comparing them as foreground colors gives a larger difference (text in
   the window RosyBrown being readable, while in terminal RosyBrown
   considerably-less-readable).


-- Stefan

Re: Some xterm-256color face colors too bright?

[David Hansen]

On Tue, 01 Feb 2005 13:48:49 +0200 stktrc <stktrc@yahoo.com> wrote:

> I have to make a reservation about sub-pixel rendering possibly
> affecting the foreground comparisons: I haven't had time to ensure
> it's off.

On Tue, 01 Feb 2005 21:52:28 +0200 Eli Zaretskii wrote:

> ??? These two paragraphs seem to contradict one another: the first one
> says that the difference between the windowed and non-windowed session
> is small (which is expected, as the former one has more colors than
> 256), while the latter says that the difference is large (``borders on
> unreadable'' vs ``everything is fine'').  I wonder how can this
> happen.  Can you shed some light on this?

Just a guess:  Xft's anti aliasing blurs the font so that it
appears brighter.

David

Re: Some xterm-256color face colors too bright?

[Stefan Monnier]

> This issue troubled me both on an LCD and an CRT display, so I assumed
> it wasn't related to anything LCD-specific such as sub-pixel
> rendering.  However, now I took the time try disabling sub-pixel
> rendering.

It's probably due to anti-aliasing.  Sub-pixel rendering is just one of the
tricks that can be used when doing anti-aliasing.

Have you tried to disable both sub-pixel rendering and anti-aliasing?

> So, it seems that the "fault" is not directly Emacs' (or so I
> speculate from the small testing I did), but the sub-pixel
> rendering's.

Or just the anti-aliasing.  What might be happening is that you're using
a very light font (at the size you're using it, its lines are of thickness
lower than 1 pixel, say), so without anti-aliasing it's "rounded up to 1",
whereas anti-aliasing actually approximates the fractional pixel
of thickness.


        Stefan

Re: Some xterm-256color face colors too bright?

[Miles Bader]

On Tue, 01 Feb 2005 23:28:09 +0200, stktrc <stktrc@yahoo.com> wrote:
> However, if that is the case I must have sub-pixel rendering enabled
> on the CRT machine, which does not sound sensible (I have assumed it
> would only be automatically enabled for LCDs).  It might be possible
> that it is enabled there anyway.

xft does both sub-pixel rendering and anti-aliasing; the former only
makes sense on LCDs, but the latter is fine for either.

The effect of either on brightness depends on the fonts of course.

To be honest this all sounds so completely
installation/setting-specific that I'm not sure there's much Emacs can
do, other than make customizing things as easy as possible.

-Miles
-- 
Do not taunt Happy Fun Ball.

Re: Some xterm-256color face colors too bright?

[Eli Zaretskii]

> From: stktrc <stktrc@yahoo.com>
> Date: Tue, 01 Feb 2005 23:28:09 +0200
> 
> The question is then, how to proceeed?  Assuming it is the sub-pixel
> rendering that causes this:
> 
> - Turn off sub-pixel rendering to gain readability for some default
>   faces in Emacs?  Would be a shame to lose all the advantages it
>   brings.
> 
> - Locally customize faces?  Doable, but what if sub-pixel rendering
>   users (= LCD display) are common, and will just increase in numbers?
>   Would be reasonable to accomodate them.
> 
> - Do nothing, because it's just me.  Do a significant amount of other
>   Emacs users experience the readability problems?
> 
> - Adjust Emacs defaults?  Is it possible without causing degradation
>   for other users?

I'd say it's something to customize locally, unless we hear similar
problems from many users.

Just to make sure it's not the color translation that causes this in
some way: if you start two Emacs sessions side by side, one windowed,
the other in xterm, and in each one say "M-x list-colors-display RET",
is there an xterm color that looks like RosyBrown, more than the color
mapped to it by default?

Re: Some xterm-256color face colors too bright?

[Eli Zaretskii]

> From: stktrc <stktrc@yahoo.com>
> Date: Wed, 02 Feb 2005 10:24:53 +0200
> 
> "Eli Zaretskii" <eliz@gnu.org> writes:
> 
> > Just to make sure it's not the color translation that causes this in
> > some way: if you start two Emacs sessions side by side, one windowed,
> > the other in xterm, and in each one say "M-x list-colors-display RET",
> > is there an xterm color that looks like RosyBrown, more than the color
> > mapped to it by default?
> 
> Terminal Emacs' list-colors-display doesn't give the same names as
> window Emacs: it has black, red etc..., color 16, color-17, ...,
> color-255.  So the color RosyBrown can't be identified there, as far
> as I can tell.

You can see to what color xterm maps RosyBrown by evaluating this
expression:

   (tty-color-desc "RosyBrown")

> But I used customize-face to set a terminal Emacs face background to
> RosyBrown and compare that with the window Emacs list-colors-display's
> RosyBrown.  They seem to be the same color.

This is expected.  So it looks like the problem is indeed in the font
used by xterm and not in the color translation.

Re: Some xterm-256color face colors too bright?

[Richard Stallman]

Would the people interested in this topic please take the discussion
off the list?