«

On Sat, 26 May 2012 09:13:35 +1200, Me <> wrote:
: On 26/05/2012 4:37 a.m., isw wrote:
:
: > It's not, but the Mac is all I know about. Have you ever used the Apple
: > "eyball" calibration (recently; it was a lot less capable in the
: > "Classic" days)? Because it sure doesn't seem to me to adjust gamma with
: > "one slider", as you claim. Each of the calibration points (maybe six or
: > seven; it's been a while) provides both brightness and color information.
: >
: No, I haven't used it for a while. But you need to understand that we
: cannot trust our eyes.
: Here are some examples from the web:
: Chromatic adaptation:
: http://www.planetperplex.com/en/item...ic-adaptation/
: Visual spatial clues throw out luminance perception:
: http://www.planetperplex.com/en/item/checker-shadow/#
: Similar image to above, but luminance perception also throws out colour
: perception:
: http://upload.wikimedia.org/wikipedi...ange_brown.svg
:
: These "tricks" are probably showing human evolutionary adaptation to
: colour perception, colour perception apparently being important to us
: for some reason (judging whether food is good and fresh etc?) as natural
: light varies so much during the day.
:
: So if you're trying to set a monitor "by eye", then you've got no
: reference point on that screen from which to make adjustments.
:
: If you've got a TV set with "scene modes", then if you change that scene
: mode from default to "movie" mode, the screen image is usually shown
: with noticeably much warmer white balance. But if you watch it for a
: few minutes, then it starts to look normal, and if you reset the TV to
: default, the standard white balance will seem to be too cool - at least
: for a short while until your eyes adjust again.
:
: When using a hardware calibration device (spyder etc) to calibrate a
: screen, it's typical when first doing the calibration to disbelieve the
: result that the calibration system is showing - as your eyes are used to
: how it was, and despite being much more colour accurate after
: calibration, it can look very wrong until your eyes adjust.
:
: That's why, within reason, unless you need to be able to match colours
: precisely across display devices - including between matching screen and
: print - then within reason, colour accuracy isn't as important as many
: people seem to think.

And anyone so committed to the ideal that he couldn't accept that reasoning
would have to allow that the color profile used in making any print must be
tuned to the ambient light in the room in which it is to be displayed. (Or the
ambient light must be tuned to color profile of the print. Either way would
work, as long as there's only one print in the room.)

Bob

Me <> wrote:

> That's why, within reason, unless you need to be able to match colours
> precisely across display devices - including between matching screen and
> print - then within reason, colour accuracy isn't as important as many
> people seem to think.

For surfing the web and office applications you're right.
If you're printing and want your prints to resemble your
monitor view, you need them calibrated, or you'll have lots
of fiddling before you.

-Wolfgang

isw <> wrote:

> I run a two-monitor setup, with a Dell 2407 (which I got because
> reviewers said it had better-than-average color quality) as the main
> screen, and my MacBook along side to provide an "auxiliary" screen.
> Although Apple displays have always been known for their better color
> performance (one reason Apple monitors have historically cost more),
> despite my best efforts I have not been able to get the laptop's screen
> to calibrate properly, nor even to have color performance that's
> remotely similar to the 2407's. A color image which straddles the two
> screens can be a painful thing to see ...

Have you profiled and calibrated (with a hardware device!) your
monitors and made sure the LUTs have been loaded into the graphic
cards and work?

-Wolfgang

On 29/05/2012 5:31 a.m., isw wrote:

>
> No; I've never been able to find out (quantitatively) why I should shell
> out the money for that. The arguments I've heard seem an awful lot like
> the ones that "golden-eared" audio folks use to convince me why vinyl
> records are "superior" to CDs -- i.e. lots of arm-waving, but never any
> numbers.
>
If you don't print, or do print but find that colour accuracy is good
enough between print and screen for your needs, then don't bother.
But your comparison above doesn't wash. If you do need to ensure colour
consistency between output devices (including print) , then you do need
to use a hardware calibrator.
Various LCD monitors I've looked at, default "out of box" colour
accuracy is typically quite poor, delta E average usually about 5 or so.
Apple computers/monitors are no exception.
There's no way to reduce that without using a hardware calibrator, as
there's no way to see "on screen" if the colour is out as you have
nothing to compare it to.

A valid comparison to Hifi "preciousness" is perhaps where people think
that it matters that they get their home cinema screens hardware
calibrated, as in that case it doesn't matter much at all if the colour
is out, and the important thing is to know how to adjust it yourself "by
eye" so that it looks good to you.

isw <> wrote:

> (Mumble) years ago, when I was designing high-resolution film printers,
> I worked very closely with a very capable photographic scientist (I did
> the electronics and optics; he did the film analysis and calibration).

> He always said that the ultimate test of a color reproduction system was
> how well it could do gray scale images -- that if there was no
> perceptible tint in the grays at any intensity from B to W, then the
> color tracking was just fine.

Truth is that our eyes are capable of detecting certain colour
casts in supposedly gray colour. And making sure that your
monitor delivers a gray gray when it should is as important today.

> If that's true (and I think that it is), then you don't need a separate
> reference.

You cannot even detect if your white is reddish or blueish
without an external reference. Is it 9000K, 6500K or 5000K?
Or something else?

If you lower the lightness somewhat you don't magically gain
an absolute eye. So the brighter grays can also be blueish
or reddish, matching to your white --- but off.

=> You need an external reference.
Since you don't want a hardware colourimeter, you need a good
reference unit (paper just doesn't cut it, especially with
optical brighteners) and a matching light (i.e. normlight)
and neutral surroundings (e.g. a normlight viewing stand).
That costs more than the colourimeter ... (and it takes a lot of
time every calibration).

So how about the brightness? Do you use your camera as a lux
meter? At factory settings, practically all flat screens are
way to bright ... and thus light areas are washed out and dark
areas are overly contrasty on such screens. Your camera is a
reference.

I already see 2 separate references ...


> And a question about hardware color calibrators: If you measure the same
> screen with several different makes of them, do they all tell you
> precisely the same thing?

Of course not. Not even the same instrument will measure the
same patch twice without some measurement error. Your camera
does have some noise, too, doesn't it? Measurement errors are
just plain physics. Photon noise is plain reality, too.


> If not, how do you know which one is correct?

If all are maintained and calibrated themselves (and use the
right matrices in case of colourimeters), their differences are
small and you can average the results. In fact, often you'll
measure important colours like "white" several times with the
same instrument to reduce the measuring error.

If one is way out, it's the wrong one.

If the photospectrometer says one thing and the tristimulus puck
says something else (and assuming the colour filters of the last
one haven't faded (the joy of plastic filters in some of them)),
the tristimulus one is using the wrong correction data (it's
blue filter doesn't perfectly match any CRTs blue phosphor ---
and with LCDs, there is a *huge* range of possible and used blues,
unlike with CRTs. Add in 'wide gamut' screens. And repeat for
red and green.).

-Wolfgang

isw <> writes:

> In fact it seems to be because the calibration procedure which works
> fine on the Dell 2407, cannot be manipulated to deliver the result that
> is necessary at several brightness steps on the MacBook's display --
> IOW, I can't manage to get the test image to appear totally neutral no
> matter what I do. I'm not sure what a hardware calibrator could do about
> that.

Inability to adjust manually doesn't prove much; our eyes are really not
very good at the task, and there isn't software really adapted to make
best use of them.

Laptops rarely have the best screens, so perhaps it won't be as good as
other things even if properly profiles; but a hardware unit that
measures the output through various filters, while driving the inputs
itself, does a MUCH better job than your eyes.
--
David Dyer-Bennet, dd-; http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info

isw <> wrote:
> Andrew Haley <> wrote:
>> isw <> wrote:

>> > And again, although I've heard people say the Apple calibration
>> > procedure is "no good", I've never been able to get any of them to
>> > elaborate on precisely which ways it fails (i.e. how do they know it's
>> > no good?).

>> Because your monitors look very different. If you'd used a hardware
>> calibrator, assuming your monitors aren't totally broken, they
>> wouldn't look very different. The black level might vary, but they'd
>> be the same brightness and the colours would be a near visual match.

> In fact it seems to be because the calibration procedure which works
> fine on the Dell 2407, cannot be manipulated to deliver the result that
> is necessary at several brightness steps on the MacBook's display --
> IOW, I can't manage to get the test image to appear totally neutral no
> matter what I do. I'm not sure what a hardware calibrator could do about
> that.

A profile allows the profile using software to turn the RGB
triplets in the image into the correct RGB values for your screen.
You cannot get the image to neutral because you have too few knobs
to describe the complexity of your monitor. A hardware method
will have 100 or more measurements and even more potential knobs,
both on and off the grey axis, to do the right thing.

You're basically trying to describe a circle with a triangle or
a rectangle --- once you have 100 corners, it's much smoother.

-Wolfgang

isw <> writes:

> In article <>,
> David Dyer-Bennet <dd-> wrote:
>
>> isw <> writes:
>>
>> > In fact it seems to be because the calibration procedure which works
>> > fine on the Dell 2407, cannot be manipulated to deliver the result that
>> > is necessary at several brightness steps on the MacBook's display --
>> > IOW, I can't manage to get the test image to appear totally neutral no
>> > matter what I do. I'm not sure what a hardware calibrator could do about
>> > that.
>>
>> Inability to adjust manually doesn't prove much; our eyes are really not
>> very good at the task, and there isn't software really adapted to make
>> best use of them
>
> But I can do a demonstrably better job of it on the Dell display. So *by
> comparison* I can't get the Mac display to look anywhere near as good. I
> think that would rule out the hardware's ability to do it better. If you
> need an "eleven" and the knob only goes to "ten", then you're just out
> of luck. A machine might break the knob trying, but you still wouldn't
> get what you need.

I'm not entirely convinced; the flaws in the displays may be different
and require different skills to correct, for example.

>> Laptops rarely have the best screens, so perhaps it won't be as good as
>> other things even if properly profiles; but a hardware unit that
>> measures the output through various filters, while driving the inputs
>> itself, does a MUCH better job than your eyes.
>
> At the risk of repeating myself, what I have been trying to find out for
> a long time (several years) is *how much better is MUCH* and *in
> precisely which ways* (Hue errors? Intensity errors? Gamma errors?, ...
> errors?)? To date, nobody has even tried to provide an answer.

Yeah, we've basically totally given up on trying to adjust displays
manually. All my experience and everything I've read about it says
that's a total dubs game, no winning condition available. Don't even
try.
--
David Dyer-Bennet, dd-; http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info

isw <> wrote:
> Andrew Haley <> wrote:
>> isw <> wrote:

>> > display -- IOW, I can't manage to get the test image to appear
>> > totally neutral no matter what I do. I'm not sure what a hardware
>> > calibrator could do about that.

>> Why do you say that? The hardware calibration procedure rewrites the
>> CLUTs in the video card to neutralize the image at all brightness
>> levels. That's the whole point of using it.

> I can certainly imagine a display so constructed (poor choice of
> backlight spectrum, poor choice of filter bandpasses) that no possible
> set of CLUT values would be "correct". Then what?

Then your manual trying would still be much worse than calibrating
with hardware. Anyway, you should dispose of such a monitor.

-Wolfgang

isw <> wrote:
> Wolfgang Weisselberg <> wrote:
>> isw <> wrote:

>> > I can certainly imagine a display so constructed (poor choice of
>> > backlight spectrum, poor choice of filter bandpasses) that no possible
>> > set of CLUT values would be "correct". Then what?

>> Then your manual trying would still be much worse than calibrating
>> with hardware. Anyway, you should dispose of such a monitor.

> I was simply observing that there could be instances when "even" a
> hardware calibrator could not produce a good result.

You were simply looking for an excuse why hardware units will
not be used by you.

-Wolfgang