Which one has the right color, Emacs or GIMP?

Which one has the right color, Emacs or GIMP?

[Lennart Borgman]

The colors I see on my screen with GIMP or Emacs for the same rgb
notation differs. For example the color #d3ff92 seems to differ quite
a bit.

What is wrong?

Re: Which one has the right color, Emacs or GIMP?

[Andreas Schwab]

Lennart Borgman <lennart.borgman@gmail.com> writes:

> The colors I see on my screen with GIMP or Emacs for the same rgb
> notation differs. For example the color #d3ff92 seems to differ quite
> a bit.
>
> What is wrong?

You should be able to double check by using GIMP's color picker.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Which one has the right color, Emacs or GIMP?

[Lennart Borgman]

On Mon, Jul 27, 2009 at 4:25 PM, Andreas Schwab<schwab@linux-m68k.org> wrote:
> Lennart Borgman <lennart.borgman@gmail.com> writes:
>
>> The colors I see on my screen with GIMP or Emacs for the same rgb
>> notation differs. For example the color #d3ff92 seems to differ quite
>> a bit.
>>
>> What is wrong?
>
> You should be able to double check by using GIMP's color picker.

Thanks. It was my typo.

Re: Which one has the right color, Emacs or GIMP?

[Jason Rumney]

Lennart Borgman wrote:
> The colors I see on my screen with GIMP or Emacs for the same rgb
> notation differs. For example the color #d3ff92 seems to differ quite
> a bit.
>
> What is wrong?
>   

GIMP, as a specialist image editing application, is probably taking 
things like monitor gamma into consideration. Emacs is just using the 
colors the default way the OS produces them, like most other applications.

Re: Which one has the right color, Emacs or GIMP?

[Stephen J. Turnbull]

Sort of OT since Lennart discovered a typo....

Jason Rumney writes:

 > GIMP, as a specialist image editing application, is probably taking 
 > things like monitor gamma into consideration.

I thought of that, but upon reconsideration I hope GIMP would not do
that.  RGBA? is the color equivalent of assembly language.

If you want to deal with monitor gamma etc, use a color space designed
for such adaptation.

Re: Which one has the right color, Emacs or GIMP?

[Miles Bader]

"Stephen J. Turnbull" <stephen@xemacs.org> writes:
>  > GIMP, as a specialist image editing application, is probably taking 
>  > things like monitor gamma into consideration.
>
> I thought of that, but upon reconsideration I hope GIMP would not do
> that.  RGBA? is the color equivalent of assembly language.
>
> If you want to deal with monitor gamma etc, use a color space designed
> for such adaptation.

Huh?  RGBA is just a notation for a storage format, it says nothing
about how the components should or should-not-be interpreted.

The gimp, however (and emacs) needs to worry about such things
regardless, since in the end it's in charge of what the user sees;
hopefully there's helpful meta-data which says what to do...

-Miles

-- 
We live, as we dream -- alone....

Re: Which one has the right color, Emacs or GIMP?

[Stephen J. Turnbull]

Miles Bader writes:

 > Huh?  RGBA is just a notation for a storage format, it says nothing
 > about how the components should or should-not-be interpreted.

Of course it says something.  It says that the various components red,
green, and blue shall be used in strengths given as a fraction of
their maximum, as modified by the alpha if the display is capable.

It's possible to quibble about whether that "strength" must be linear
or merely increasing (the latter interpretation admits gamma
adjustment among others).  But to say "it says nothing" is too much of
a stretch.

I admit I've never read the X Window System Color Management System
documents closely, but I suspect there's a specification there for RGB.

I can say that one of the rationales for RGB.txt was so that color
names could be device-independent, while their RGB values might vary.
But of course that didn't work so well because RGB.txt lives where
libX11 does, not on the server, and nowadays RGB.txt may not be
consulted at all.

 > The gimp, however (and emacs) needs to worry about such things
 > regardless, since in the end it's in charge of what the user sees;
 > hopefully there's helpful meta-data which says what to do...

IMHO, Emacs should delegate this to either the display hardware and
drivers (which should provide a satsifactory default gamma
adjustment), or to another application such as the GIMP.

Re: Which one has the right color, Emacs or GIMP?

[Miles Bader]

"Stephen J. Turnbull" <stephen@xemacs.org> writes:
>  > Huh?  RGBA is just a notation for a storage format, it says nothing
>  > about how the components should or should-not-be interpreted.
>
> Of course it says something.  It says that the various components red,
> green, and blue shall be used in strengths given as a fraction of
> their maximum, as modified by the alpha if the display is capable.
>
> It's possible to quibble about whether that "strength" must be linear
> or merely increasing (the latter interpretation admits gamma
> adjustment among others).  But to say "it says nothing" is too much of
> a stretch.

Er, sure as a vague "the components have a positive correlation with the
redness, greenness, and blueness of the resulting color the user sees"
notion, sure, there's some information there. :)

I'm sure there's some at least _somewhat_ more precise definition
(linear/exponential/whatever, though I suspect not much more) for things
sent to the X server and stored in rgb.txt.

> IMHO, Emacs should delegate this to either the display hardware and
> drivers

Sadly, they're usually all fucked up too though.

In general, Emacs _can't_ really just ignore the issue unless it's
merely acting as a middleman exchanging colors amongst components which
are known to be in agreement about their interpretation.  Since Emacs is
_not_ just doing that -- it also reads image files using a variety of
libraries which may not have compatible interpretations for the values
they yield, and [pertinently to this thread] uses color numbers from the
user.  Since the latter are presumably intended to be at least somewhat
platform independent, and different platforms can have wildly different
display systems, we then have to worry about the issue for those.

-Miles

-- 
It wasn't the Exxon Valdez captain's driving that caused the Alaskan oil spill.
It was yours.  [Greenpeace advertisement, New York Times, 25 February 1990]

Re: Which one has the right color, Emacs or GIMP?

[Stephen J. Turnbull]

Miles Bader writes:

 > > IMHO, Emacs should delegate this to either the display hardware and
 > > drivers
 > 
 > Sadly, they're usually all fucked up too though.

In which case, how is Emacs supposed to know what to do?  AFAIK there
is no such metadata for most formats (TIFF probably has it, Postscript
too I suppose, but surely not GIF or BMP), and if the hardware is not
reliable, how likely is it that it provides "I'm fucked up and this is
precisely how" metadata? :-)

 > _not_ just doing that -- it also reads image files using a variety of
 > libraries which may not have compatible interpretations for the values

And for which the metadata is missing....

 > they yield, and [pertinently to this thread] uses color numbers from the
 > user.

Which sounds like a perfectly good application for an (optional) Lisp
library that the user invokes at need.

 > Since the latter are presumably intended to be at least somewhat
 > platform independent, and different platforms can have wildly different
 > display systems, we then have to worry about the issue for those.

Well, no, you don't, since it was a pilot error in the first place.
And we're back to user assistance libraries.

Re: Which one has the right color, Emacs or GIMP?

[Miles Bader]

"Stephen J. Turnbull" <stephen@xemacs.org> writes:
>  > _not_ just doing that -- it also reads image files using a variety of
>  > libraries which may not have compatible interpretations for the values
>
> And for which the metadata is missing....

Sadly it happens often, of course...

Then you either have to decide to use a common default and hope maybe
it's right/close, or tell the user "all bets are off if you use images
like that."  Not that the user is going to forgive Emacs if it guesses
differently than other popular apps (firefox or whatever)...

>  > they yield, and [pertinently to this thread] uses color numbers from the
>  > user.
>
> Which sounds like a perfectly good application for an (optional) Lisp
> library that the user invokes at need.

Sure, but then the _library_ needs to be know what it's targeting, which
requires emacs to either commit to something, or have a mechanism for
telling said library what interpretation the rendering system provides
(if any...).  So while it might make some Emacs core code easier, it
doesn't really resolve Emacs of responsibility for caring at least a
little bit.

Of course, the as Emacs starts to do more than just shuffling colors
around -- if it adjusts color/image intensities (it already does,
actually, but only a little) or starts doing blending of colors (in the
presence of alpha information) -- then even the Emacs core code has to
at least think about the issues.

Emacs core code does try to deal with gamma correction now, though I
dunno how correct any of that code is...

ugh...

-miles

-- 
We live, as we dream -- alone....