Color profiles and correct usage

[Jo_y]

Hm, i'm confused about Color Profiles. And i'm for sure not the first
one with this topic. (Its always a mess between different devices)
Till today I used a relative good color balance while printing, but
now, I want to improve the quality of my prints.

I own a:
-Epson stylus dx4000
-Mac os 10.6
-Preview (to print images)
-Gimp, fewer Photoshop

I used to use:
Screen: Apple RGB (it seems cold, but it has a really nice balance
compared to the firmware profile)
Painting/ photo retouching: Generic RGB (should be the common profile
to display images within a image processor)
image-file-profile: sRGB to reinforce colors

my prints (on normal write paper, 297 x 210 mm) look instead:
a saturated orange changes in a yellowish tone
a dark blue changes in a lilla-like tone
the values are absolute less saturated

Maybe I've to print my pictures from Scribus to get clear results ?
At last, I need to know if values like perceptual, saturation,
absolute/relative colorimetric do influence heavily the output print,
or not.

[Rob]

Also you could post to comp.periphs.printers newsgroup.

each type of paper and ink has its own profile combination. you should
download and try the profile for that paper which you are using and
compare the difference.

Epson have a setup PDF document to download, for photoshop which you may
like to read.

called RGB Workflow - Photoshop CS4 / Win & MAC (PDF) thats the same for
CS5.

[Me]

Missing piece of equipment is /calibrated/ display if you want accurate
sceen:print matches. That probably means investing in a hardware
colorimeter package. The software calibration methods on MAc and
Windows are pretty useless, but you can check the state of your display
here:
http://www.lagom.nl/lcd-test/
The mis-match mentioned by the OP is probably because of a poorly
calibrated display, as using the printer driver to manage colour,
(rather than a colour managed process using ICC profiles) usually gives
a pretty decent result with Epson printers/papers. Another possibility
is incorrect paper or printer driver settings.
Part of the coloiur managed process is to allow soft-proofing, viewing
the image on screen with the ICC profile for the printer/paper
combination loaded, and making minor pre-print adjustments from there,
But you need to get the monitor calibration nailed for there to be any
real point in doing this.

[Andrew Haley]

In addition to all that, saturated orange is nearly impossible to
print. A good profile will be acurate, but it can't help the printer
to print out of gamut colours.

Andrew.

[Rob]

>> called RGB Workflow - Photoshop CS4 / Win& MAC (PDF) thats the same for

>> CS5.
>
> In addition to all that, saturated orange is nearly impossible to
> print. A good profile will be acurate, but it can't help the printer
> to print out of gamut colours.
>
> Andrew.

Orange and green are easy to print on the Epson 7900 :)

[isw]

In article <jpl35q$ajj$1...@news.albasani.net>, Me <us...@example.net>
wrote:

My Dell 2407WFPHC monitor is calibrated using the built-in Mac method
you call "pretty useless". A visit to the site you reference (I've been
there before) shows that everything looks just as the author says it
should, for a properly adjusted display.

So could you please elaborate on precisely which ways Apple's
calibration method is "useless"? I'm sincerely curious as to what that
method's faults are, and how much better my monitor's calibration could
be.

Isaac

[Me]

>>> called RGB Workflow - Photoshop CS4 / Win& MAC (PDF) thats the same for

>>> CS5.
>>>
>> Missing piece of equipment is /calibrated/ display if you want accurate
>> sceen:print matches. That probably means investing in a hardware
>> colorimeter package. The software calibration methods on MAc and
>> Windows are pretty useless, but you can check the state of your display
>> here:
>> http://www.lagom.nl/lcd-test/
>
> My Dell 2407WFPHC monitor is calibrated using the built-in Mac method
> you call "pretty useless". A visit to the site you reference (I've been
> there before) shows that everything looks just as the author says it
> should, for a properly adjusted display.
>
> So could you please elaborate on precisely which ways Apple's
> calibration method is "useless"? I'm sincerely curious as to what that
> method's faults are, and how much better my monitor's calibration could
> be.
>
> Isaac

For general use, software calibration is fine. However, where you need
accurate colour match between output devices, it's useless.
An example is using the software gamma charts, where sharpness setting
on the display has a massive effect on perceived sharpness. So you can
use a sharpening chart to set that - as on the lagom site - but that's
useless because (conversely) the gamma setting will affect the perceived
sharpness on that chart. On this basis, which should you set first "by
eye". The answer is neither - you need to use hardware calibration, or
you are running around in circles. You could think that by looking at
the lagom images for sharpness and gamma that you've got both right, but
chances are you've got both wrong. You won't get it right, because the
gamma setting in the software usually has only one slider to adjust a
curve, and takes no account of likely non-linear response of the
display, so you're basically stuffed if setting gamma for different
luminance levels - then throw in R,G, & B, and it's a big pile of worms.
It's almost completely futile using a colour managed printing process
with a screen display that isn't properly hardware calibrated. At best
you could load ICC profiles in to PS, check gamut warnings etc, but if
you don't get a perfect screen:print match, then you're none the wiser
as to whether the printer is the cause, the ICC profile is wrong (not
totally unlikely), or the monitor is wrong.
This (I hope) isn't a mac vs windows discussion. Macs and windows are
about the same for colour management, and typical mac hardware is not
anything special - nor bad - for graphics.

[Me]

> on the display has a massive effect on perceived <correction> - "gamma". So you can

[nospam]

In article <isw-00CC64.23284324052012@[216.168.3.50]>, isw

<i...@witzend.com> wrote:

> > Missing piece of equipment is /calibrated/ display if you want accurate
> > sceen:print matches. That probably means investing in a hardware
> > colorimeter package. The software calibration methods on MAc and
> > Windows are pretty useless, but you can check the state of your display
> > here:
> > http://www.lagom.nl/lcd-test/
>
> My Dell 2407WFPHC monitor is calibrated using the built-in Mac method
> you call "pretty useless". A visit to the site you reference (I've been
> there before) shows that everything looks just as the author says it
> should, for a properly adjusted display.
>
> So could you please elaborate on precisely which ways Apple's
> calibration method is "useless"? I'm sincerely curious as to what that
> method's faults are, and how much better my monitor's calibration could
> be.

because it's done by eye, not hardware. the eye is not accurate enough.

[isw]

In article <250520120927401256%nos...@nospam.invalid>,

So then why does my display look just like that site says it should? How
far off is an "eyeball" calibration, compared to an instrumented one?

(IOW, how much better will it be if I lay out a couple hundred bucks for
a calibrator -- 2%, or 20%?)

Isaac

[isw]

In article <jpnc8a$l0c$1...@news.albasani.net>, Me <us...@example.net>

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.

Isaac

[nospam]

In article <isw-773AD4.09292225052012@[216.168.3.50]>, isw

<i...@witzend.com> wrote:

> > > So could you please elaborate on precisely which ways Apple's
> > > calibration method is "useless"? I'm sincerely curious as to what that
> > > method's faults are, and how much better my monitor's calibration could
> > > be.
> >
> > because it's done by eye, not hardware. the eye is not accurate enough.
>
> So then why does my display look just like that site says it should? How
> far off is an "eyeball" calibration, compared to an instrumented one?

it can potentially be quite a bit, but on the other hand, you could be
lucky and hit it close enough. your eye is not accurate and easily
fooled.

<http://blogs.discovermagazine.com/badastronomy/2009/06/24/the-blue-and-
the-green/>

run the software calibrator and move your head and see how much it
varies. change the ambient lighting too. a hardware puck is immune to
that.

> (IOW, how much better will it be if I lay out a couple hundred bucks for
> a calibrator -- 2%, or 20%?)

they're under $100 and you might even be able to rent one for even less.

[Savageduck]

I use the Pantone huey Pro, $99, on my Mac.
< http://www.pantone.com/pages/products/product.aspx?pid=562&ca=2 >


--
Regards,

Savageduck

[Me]

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/chromatic-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/wikipedia/commons/9/9a/Optical_grey_squares_orange_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.

[isw]

In article <250520121243288841%nos...@nospam.invalid>,

nospam <nos...@nospam.invalid> wrote:

> In article <isw-773AD4.09292225052012@[216.168.3.50]>, isw
> <i...@witzend.com> wrote:
>
> > > > So could you please elaborate on precisely which ways Apple's
> > > > calibration method is "useless"? I'm sincerely curious as to what that
> > > > method's faults are, and how much better my monitor's calibration could
> > > > be.
> > >
> > > because it's done by eye, not hardware. the eye is not accurate enough.
> >
> > So then why does my display look just like that site says it should? How
> > far off is an "eyeball" calibration, compared to an instrumented one?
>
> it can potentially be quite a bit, but on the other hand, you could be
> lucky and hit it close enough. your eye is not accurate and easily
> fooled.
>
> <http://blogs.discovermagazine.com/badastronomy/2009/06/24/the-blue-and-
> the-green/>
>
> run the software calibrator and move your head and see how much it
> varies. change the ambient lighting too. a hardware puck is immune to
> that.

The puck is "immune" simply because it cannot move. That means that it's
measurements will only be valid if I view the screen at precisely the
same angle it did. Anyhow, when I do photo editing, I'm careful to sit
"front and center" to the screen.

> > (IOW, how much better will it be if I lay out a couple hundred bucks for
> > a calibrator -- 2%, or 20%?)
>
> they're under $100 and you might even be able to rent one for even less.

But you still haven't offered any actual explanation of why I'd want to.
You spoke with authority, claiming that visual calibration was no good.
How do you know that? What is your evidence? How much better would
"machine" calibration actually be?

Isaac

[isw]

In article <jposlv$cq5$1...@news.albasani.net>, Me <us...@example.net>
wrote:

Now, that statement is something I can totally agree with.

Isaac

[Andrew Haley]

Well, it saves a lot of paper; all that printing of test images gets
tedious before very long. And also, if someone else is doing the
printing, there's a lot to be said for having a good idea of how the
print might come out. I suppose there are some photographers who
don't print, but even then I don't suppose they look at only one monitor.

Andrew.

[Eric Stevens]

On Fri, 25 May 2012 20:32:37 -0700, isw <i...@witzend.com> wrote:

>But you still haven't offered any actual explanation of why I'd want to.
>You spoke with authority, claiming that visual calibration was no good.
>How do you know that? What is your evidence? How much better would
>"machine" calibration actually be?

When I first took up digital photography I struggled to get prints
from my Epson Stylus 1890 (I think it was, then) which were even
pleasant to look at. Over a period of a year I tried all kinds of ICC
profiles and used a vast quantity of expensive paper in test prints. I
suspected my problem was the match between printer and screen and used
a variety of software applications which told me my screen calibration
was pretty close. But still I couldn't produce prints which looked at
all like the images I saw on the screen.

Eventually I bit the bullet and bought a Spyder. This caused an
immediate, but subtle, change in the screen calibration. Only a little
more fiddling after this was required to get prints (now from my Epson
3900) which are so close to screen image that I don't even bother to
'proof' them on the screen. My experience is such that I wouldn't go
back to the Mk 1 eyeball. Continued reliance on this grade of
technology is a recipe for frustration and unsatisfactory prints.

[Robert Coe]

On Sat, 26 May 2012 03:00:14 -0500, Andrew Haley
<andr...@littlepinkcloud.invalid> wrote:

They don't, and my laptop displays colors in a way that's annoyingly different
from my monitors at home and at work. So when I edit images on the train to or
from work, I have to remind myself to stick to brightness and composition
issues and not get too finicky about color values, which I'd just have to
re-jigger later.

Bob

[Robert Coe]

On Sat, 26 May 2012 09:13:35 +1200, Me <us...@example.net> wrote:

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

[isw]

In article <fid2s7ld13rui3sqe...@4ax.com>,

For all of their faults, one advantage of CRT technology was that,
because of the very limited set of phosphors manufacturers had available
to choose from, most of them exhibited rather similar colorimetry.

LCD technology changed all that. Not only can the backlight's spectrum
be chosen almost arbitrarily, but there's essentially no constraint on
the colors that can be used for the three primary filters,since they are
just transparent colored dyes. The result is that a flat panel display's
characteristics can just as easily be chosen for low manufacturing cost,
high brightness, long life, lower battery drain, and/or better stability
against fading, than, say, best color reproduction. Assuming that the
manufacturer's engineers were even aware of the desirability of accurate
color reproduction in the first place, much less were told to make that
a design goal, of course.

While it is possible to find LCD panels that have decent colorimetry,
they tend to be somewhat more expensive. I'd bet that a "standard"
display in a "standard" laptop wouldn't be in that category.

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 ...

Isaac

[Andrew Haley]

isw <i...@witzend.com> wrote:
> In article <fid2s7ld13rui3sqe...@4ax.com>,
> Robert Coe <b...@1776.COM> wrote:
>
>> On Sat, 26 May 2012 03:00:14 -0500, Andrew Haley
>> <andr...@littlepinkcloud.invalid> wrote:
>> :

>> : Well, it saves a lot of paper; all that printing of test images
>> : gets tedious before very long. And also, if someone else is
>> : doing the printing, there's a lot to be said for having a good
>> : idea of how the print might come out. I suppose there are some
>> : photographers who don't print, but even then I don't suppose they
>> : look at only one monitor.
>>
>> They don't, and my laptop displays colors in a way that's
>> annoyingly different from my monitors at home and at work. So when
>> I edit images on the train to or from work, I have to remind myself
>> to stick to brightness and composition issues and not get too
>> finicky about color values, which I'd just have to re-jigger later.
>

> 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.

I wouldn't expect a perfect visual match, but you should be able to
get very close with any decent calibrator. Most laptops have a
terrible problem with viewing angle, though, and that can't be fixed.

Andrew.

[Wolfgang Weisselberg]

isw <i...@witzend.com> 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

[Wolfgang Weisselberg]

Me <us...@example.net> 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]

In article <ghga99-...@ID-52418.user.berlin.de>,

Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

> isw <i...@witzend.com> 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

Yes.

> (with a hardware device!) your monitors

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.

>and made sure the LUTs have been loaded into the graphic
> cards and work?

Yes. I used the Apple-provided calibration procedure on both monitors.
That app can maintain a list of calibrations, and I can switch between
them and see the difference.

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?).

Isaac

[Me]

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]

In article <jq1a32$crs$1...@news.albasani.net>, Me <us...@example.net>
wrote:

> 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.

(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.

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

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

Isaac

[Wolfgang Weisselberg]

isw <i...@witzend.com> 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

[Andrew Haley]

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.

Andrew.

[isw]

In article <jcGdnV2XnaYlbFXS...@supernews.com>,

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.

The MacBook came with an Apple-supplied profile for that display; it
never looked good, even when the Mac was new. It looks like it was
intended to maximize brightness with "tolerable" color accuracy.

Isaac

[Andrew Haley]

isw <i...@witzend.com> wrote:
> In article <jcGdnV2XnaYlbFXS...@supernews.com>,
> Andrew Haley <andr...@littlepinkcloud.invalid> wrote:
>
>> isw <i...@witzend.com> 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.

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.

Andrew.

[David Dyer-Bennet]

isw <i...@witzend.com> 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...@dd-b.net; 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]

In article <ylfklik5...@dd-b.net>,

David Dyer-Bennet <dd...@dd-b.net> wrote:

> isw <i...@witzend.com> 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.

> 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.

An audio comparison might be "Amplifier A exhibits 5% harmonic
distortion, while amplifier B has only 0.1% under similar conditions".
Saying that "Amplifier A (or B) sounds MUCH better" tells me exactly
nothing -- especially when it is known that some sorts of distortion
make audio sound "better" (so-called "euphonic" distortion), which is
probably why some folks prefer vinyl discs. I would not be surprised to
learn that euphonic (euvisic??) distortions are present in visual
situations, too.

But anyway, if the difference between the two calibration methods can't
be expressed with numbers, then I'm siding with Lord Kelvin.

Isaac

[isw]

In article <Z-2dnUShxJZ_UFTS...@supernews.com>,

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?

Isaac

[David Dyer-Bennet]

isw <i...@witzend.com> writes:

> In article <ylfklik5...@dd-b.net>,
> David Dyer-Bennet <dd...@dd-b.net> wrote:
>
>> isw <i...@witzend.com> 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.

[Wolfgang Weisselberg]

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

[Andrew Haley]

I think that falls in the category "totally broken," with the caveats
above. AFAIK that's pretty unusual. The most likely result would be
that you'd have a very poor gamut and/or brightness level.

Andrew.

[Andrew Haley]

isw <i...@witzend.com> wrote:
> In article <ylfklik5...@dd-b.net>,
> David Dyer-Bennet <dd...@dd-b.net> wrote:
>
>> isw <i...@witzend.com> 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.

Maybe contrast boost is turned on. You can find it under System
Preferences>Universal Access. Then look under the "Seeing" Tab, and
then under "Display." Next, move the "Enhance Contrast" slider all the
way to the left.

>> 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.
>
>> 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.

The equivalent would be Delta E errors. No-one can really tell you
how good your particular monitor could be without trying it. A good
monitor for colour work might have an average Delta E of 1ish and a
worst case of of 3ish. 8-bit CLUTs, which is what you usually get,
are a bit too crude to get reference-quality results.

Tom's Hardware reckons that a MacBook can be calibrated to have a
average Delte E of 3.19.

Andrew.

http://www.tomshardware.com/reviews/apple-macbook-laptop,2130-3.html

[Wolfgang Weisselberg]

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

-Wolfgang

[isw]

In article <jaju99-...@ID-52418.user.berlin.de>,

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

Isaac

[isw]

In article <UbKdnfPwTpWedVHS...@supernews.com>,

Andrew Haley <andr...@littlepinkcloud.invalid> wrote:

> isw <i...@witzend.com> wrote:
> > In article <ylfklik5...@dd-b.net>,
> > David Dyer-Bennet <dd...@dd-b.net> wrote:
> >
> >> isw <i...@witzend.com> 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.
>
> Maybe contrast boost is turned on. You can find it under System
> Preferences>Universal Access. Then look under the "Seeing" Tab, and
> then under "Display." Next, move the "Enhance Contrast" slider all the
> way to the left.

It's not on. And anyway, if it were, it would affect both monitors. I
can calibrate the Dell just fine.

Isaac

[isw]

In article <diaq99-...@ID-52418.user.berlin.de>,

It sounds like you know for a fact that the Apple "eyeball" calibration
method has only a limited ability to manipulate control points, compared
to a hardware calibrator. Is that really the case? I don't know of any
reason why it would have to be so.

Isaac

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:

> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

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

> It sounds like you know for a fact that the Apple "eyeball" calibration
> method has only a limited ability to manipulate control points, compared
> to a hardware calibrator. Is that really the case? I don't know of any
> reason why it would have to be so.

According to
http://www.youtube.com/watch?v=51gbxDwz0NM
there are only 5 grey/gamma points.

Thus, only 5 control points. Contrast and compare to 100,
500, 836 or more control points.

Yes, 5 control points, at 6, 12.5, 25, 50 and 75% grey may
have been enough for a well behaved CRT[1]. Well, not really.

For a LCD that may not enough[4].

VMRL (.wrl) files on request. I can also run a different 836
(or more, or less) patches check (hitting different spots)
to evaluate the capability of a L*a*b* clut profile from the
original 836 patches, if you're interested.

-Wolfgang

[1] Checked my CRT with a photospectrometer[2]: 0% Black, 100%
White, Red, Green, Blue and the relevant 6%, 12.5%, 25%, 50%
and 75% grey. Created a Gamma+Matrix profile from the data.
Checked the profile against the full 800+ control points[3]
from which these 10 values were pulled. Delta E (CIEDE2000):
RMS and average <1 (probably invisible), maximum >3 however
(clearly visible), and the plot shows that all the dark
colours are well off, due to non-linearity in all 3 colours
in the dark area.

[2] Which delivers absolute values and full spectral data ---
quite unlike your eye.

[3] 85 steps on the gray axis, 65 steps on each colour axis,
the rest well distributed.

[4] Computed as in [1], laptop LCD (TN) screen, white LEDs.
Delta E (CIEDE2000): average >3.6, RMS >4.0, maximum >8.0.
Very bad. Everything's visibly off (even on the gray axis!).
The plot shows most areas are off, only one part of the grey
axis is sorta neutral.

Retry with a Shaper+Matrix profile. Slightly better average
and RMS, maximum DE error now >11. The plot shows the grey
axis being mostly correct in the darkest part and returning
to neutral in 2 or 3 other places.

[Wolfgang Weisselberg]

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

-Wolfgang

[Eric Stevens]

On Thu, 7 Jun 2012 13:51:45 +0200, Wolfgang Weisselberg
<ozcv...@sneakemail.com> wrote:

>isw <i...@witzend.com> wrote:
>> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>
>>> You're basically trying to describe a circle with a triangle or
>>> a rectangle --- once you have 100 corners, it's much smoother.
>
>> It sounds like you know for a fact that the Apple "eyeball" calibration
>> method has only a limited ability to manipulate control points, compared
>> to a hardware calibrator. Is that really the case? I don't know of any
>> reason why it would have to be so.
>
>According to
>http://www.youtube.com/watch?v=51gbxDwz0NM
>there are only 5 grey/gamma points.

There is nothing absolute about a colour management process which
enables you to "choose your display to be as sharp and vivid as you
prefer it to be". It might be OK if all you want to do is stare at the
screen but not if you want to print to a colour-managed printer.

Here is how to do it with the Spyder 2 (now they are up to Spyder 4)
http://www.youtube.com/watch?v=OU6T3fB2gVg&feature=related

Regards,

Eric Stevens

[Wolfgang Weisselberg]

Eric Stevens <eric.s...@sum.co.nz> wrote:
> On Thu, 7 Jun 2012 13:51:45 +0200, Wolfgang Weisselberg

>>isw <i...@witzend.com> wrote:
>>> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

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

>>> It sounds like you know for a fact that the Apple "eyeball" calibration
>>> method has only a limited ability to manipulate control points, compared
>>> to a hardware calibrator. Is that really the case? I don't know of any
>>> reason why it would have to be so.

>>According to
>>http://www.youtube.com/watch?v=51gbxDwz0NM
>>there are only 5 grey/gamma points.

> There is nothing absolute about a colour management process which
> enables you to "choose your display to be as sharp and vivid as you
> prefer it to be".

With any colour management process your're free to detune your
display and use sharpening and vibrancy settings as you like.

> It might be OK if all you want to do is stare at the
> screen but not if you want to print to a colour-managed printer.

Not with 5 point adjustments.

-Wolfgang

[isw]

In article <lvgca9-...@ID-52418.user.berlin.de>,

I don't agree -- it depends on how much interpolation can be done from
those points, and I suspect that it could result in a pretty decent
curve. The interpolation should work well, because it is a
"well-behaved" function.

But anyhow, if it's not possible for a human to create a decent match
*at those points*, why do you think that a machine could do it? As I
said, if you need an "eleven" and the knob only goes to "ten" ...

Isaac

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:
> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>> Eric Stevens <eric.s...@sum.co.nz> wrote:
>> > On Thu, 7 Jun 2012 13:51:45 +0200, Wolfgang Weisselberg
>> >>isw <i...@witzend.com> wrote:
>> >>> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

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

>> >>> It sounds like you know for a fact that the Apple "eyeball" calibration
>> >>> method has only a limited ability to manipulate control points, compared
>> >>> to a hardware calibrator. Is that really the case? I don't know of any
>> >>> reason why it would have to be so.

>> >>According to
>> >>http://www.youtube.com/watch?v=51gbxDwz0NM
>> >>there are only 5 grey/gamma points.

>> > There is nothing absolute about a colour management process which
>> > enables you to "choose your display to be as sharp and vivid as you
>> > prefer it to be".

>> With any colour management process your're free to detune your
>> display and use sharpening and vibrancy settings as you like.

>> > It might be OK if all you want to do is stare at the
>> > screen but not if you want to print to a colour-managed printer.

>> Not with 5 point adjustments.

> I don't agree --

Fine. Do you have data? Facts?

> it depends on how much interpolation can be done from
> those points,

Oh, you can interpolate as much as you like. The question
is: how close do you come to reality?

> and I suspect that it could result in a pretty decent
> curve.

Sure. A broken clock could result in showing the right time,
too. Happens pretty regularly, twice a day, in fact.

The idea is not to "suspect" a pretty decent result, but to get
one, every time.

> The interpolation should work well, because it is a
> "well-behaved" function.

Do you have data for that? Do you have facts?

> But anyhow, if it's not possible for a human to create a decent match
> *at those points*, why do you think that a machine could do it?

Why not?
Humans cannot fly ... but with machines they can.
Humans cannot see single cells ... but with microscopes they can.
Humans cannot listen to radio waves ... but with radios they can.

Humans cannot measure brightness and saturation by eye with
high accuracy nor absolutely ... but with machines they can.

Still think not?

> As I
> said, if you need an "eleven" and the knob only goes to "ten" ...

Oh, but the problem is you cannot even see the knob properly
without a machine ... much less see what it is set to.

-Wolfgang

[isw]

In article <j1kfa9-...@ID-52418.user.berlin.de>,

I'm still trying to find someone who actually *knows* in precisely what
ways the Apple "eyeball" calibration method is flawed. Because from my
understanding of how it works, it should produce pretty decent results.

Your statements and questions above make it pretty clear that you're not
going to be the one who provides the answer.

Isaac

[Eric Stevens]

It wsill produce pretty decent results if all you are after is that it
looks OK. If you want to make sure that when you edit an image its
colours on the screen match the colours that your printer is going to
produce then it will be about as adequate as navigating by 'going a
long way, that-away (points)'. As Edward Demming didn't actually say
"You can't manage what you can't measure."

>
>Your statements and questions above make it pretty clear that you're not
>going to be the one who provides the answer.
>

... and I suspect you are not going to be the one to recognise it when
he does. :-)

Regards,

Eric Stevens

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:

> I'm still trying to find someone who actually *knows* in precisely what
> ways the Apple "eyeball" calibration method is flawed.

I gave you the answer a couple posts ago.
Short form: 5 measurement points on the grey axis aren't enough
for many screens even *IF* the eye was a perfect instrument.

Delta E (2000) averages of 4.0 and maximums of 8 are BAD.

FYI:
ΔE means
0.0 … 0.5 practically no difference
0.5 … 1.0 might be visible to the trained eye
1.0 … 2.0 visible difference if you look closely
2.0 … 4.0 clear colour differences
4.0 … 5.0 rarely tolerated colour differences
more than 5.0 that's a completely different colour

So if your screen, on the average, is showing hard to tolerate
colour shifts and for some colours even completely different
colours ... then the calibration method is unusable.

Now add in that your eye is not a perfect instrument. That it
*will* be off by some degree for every measured point by about
1 ∆E or more. (If you want, I can randomize the measured data
for the 5 points by 1-2 ∆E and compute the result based on that.)

> Because from my
> understanding of how it works, it should produce pretty decent results.

Decent as in 'the grey axis will look quite neutral to the
untrained eye, and the colours will be similar', yes.

Decent as in 'If you hold your print next to your monitor it'll
look the same[1][2]', not in a 1000 years.

Decent as in 'The monitor will give you a good idea how the print
will look like', depends a lot on how good an idea you want.

If you've not got a good eye for colour and look at the print
a few minutes after you've seen the monitor and don't have any
comparisons when looking at the print (i.e. you're not trying to
hit a specific colour, say a company logo's colour or a specific
flower and thus are not holding the print next to such an object),
you will be OK.

> Your statements and questions above make it pretty clear that you're not
> going to be the one who provides the answer.

It seems you're not even getting that you have been handed facts.
Or at least that you have problems evaluating what they mean.
OR that you are a fanboi and therefore there are no arguments
that could make you change your mind, since that would challenge
a dogma or something.

If it's the first two, feel free to ask "what does ∆E mean?"
or "How do you arrive at these ∆E numbers?" or similar.

Here:
https://gist.github.com/2929806
is the 836-patches file (Laptop.ti3), the extracted 5-grey-patches
file (5-Laptop.ti3) and the error that happens when you turn
5-Laptop.ti3 into an ICC (matrix+shaper) and test it with
Laptop.ti3 (5-Laptop.wrl) (colour coded by ∆E).

Feel free to play with the values, feel free to experiment with
other ICC generation methods, feel free to report back if you
manage to get a better result (and if so, how).

-Wolfgang

[1] assuming all colours are in the gamut of the print and of
the monitor
[2] you also should use controlled lighting and have the
print corrected for the lighting used (by default D50)
and the monitor set to emulate for the same lighting

[isw]

In article <ptvoa9-...@ID-52418.user.berlin.de>,

Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

> isw <i...@witzend.com> wrote:
>
> > I'm still trying to find someone who actually *knows* in precisely what
> > ways the Apple "eyeball" calibration method is flawed.
>
> I gave you the answer a couple posts ago.
> Short form: 5 measurement points on the grey axis aren't enough
> for many screens even *IF* the eye was a perfect instrument.

The human eye is very good at making comparisons and determining when
two things are the same -- which is the way the Apple method works. Have
you ever used it?

> So if your screen, on the average, is showing hard to tolerate
> colour shifts and for some colours even completely different
> colours ... then the calibration method is unusable.

Or (as I suspect might be the case) the screen (and its driving
hardware) is simply not able to be calibrated properly.

Using the Apple calibration method, I can do a perfectly decent job on
the Dell 2407; all the test images on the web site you referred to look
just fine. I cannot achieve similar performance from the MacBook's
display, whether I use the Apple-provided profile, or one of my own --
though the latter are (I think) somewhat better.

I have, on several occasions, tried to calibrate other LCD screens as
well, also mostly with poor results. It is known that LCD screens show
much more variability in color accuracy than CRTs (usually), and that
would seem to indicate that not all screens can be properly calibrated.
Otherwise, why would it even be mentioned in reviews? I bought the 2407
after learning that it did, indeed, have good color performance.

> > Because from my
> > understanding of how it works, it should produce pretty decent results.
>
> Decent as in 'the grey axis will look quite neutral to the
> untrained eye, and the colours will be similar', yes.

Well, the Dell 2407 is, and the MacBook is not, capable of producing a
neutral grey scale; that is precisely the problem. It cannot do that
even at the five points that you claim are insufficient. I have a
feeling that if (as you claim) a hardware calibrator can manage to stick
values in the LUT that the "eyeball" method cannot, the result would be
better color accuracy at the expense of some other parameter, possibly
brightness or contrast ratio. Given that choice I (and perhaps Apple)
would choose a bright display. After all, the Mac is a portable
computer, not a dedicated photo editor; the screen needs to be usable
under a wide range of situations.

> Decent as in 'If you hold your print next to your monitor it'll
> look the same[1][2]', not in a 1000 years.

FWIW, I'm not interested in hard-copy output; it really is pretty
limited, both in gamut, and in contrast ratio as well.

> Decent as in 'The monitor will give you a good idea how the print
> will look like', depends a lot on how good an idea you want.
>
> If you've not got a good eye for colour and look at the print
> a few minutes after you've seen the monitor and don't have any
> comparisons when looking at the print (i.e. you're not trying to
> hit a specific colour, say a company logo's colour or a specific
> flower and thus are not holding the print next to such an object),
> you will be OK.
>
> > Your statements and questions above make it pretty clear that you're not
> > going to be the one who provides the answer.
>
> It seems you're not even getting that you have been handed facts.
> Or at least that you have problems evaluating what they mean.
> OR that you are a fanboi and therefore there are no arguments
> that could make you change your mind, since that would challenge
> a dogma or something.

If I were a "fanboi", would I be talking about how good the Dell display is, and
complaining about the one in the Mac?

I just want some facts, specifically explaining *how* the Apple method
is flawed. All you've done is to say over and over that the mechanized
method is better. I want to know *why*. Saying that "you get more, and
more precise, numbers" doesn't mean anything unless you can state
precisely *why* that difference matters. It's like the audio folks who
go on about 24 bit samples at 96 or 192 kilosamples/second, and how much
"better" that is than 16 bit, 44.1 kilosample sound. Sure, the numbers
are bigger, but in fact, the difference is simply not audible, so who
cares?

> If it's the first two, feel free to ask "what does ΔE mean?"
> or "How do you arrive at these ΔE numbers?" or similar.

I have some knowledge of those, and a few other, things concerning color
imaging. Prior to retiring, I spent many years in television technology
including some work with very large screen projection systems, helped
develop MPEG-2, and designed a few high-performance film printers, among
other things.

Isaac

[nospam]

In article <isw-0B81EF.00344215062012@[216.168.3.50]>, isw

<i...@witzend.com> wrote:

> > > I'm still trying to find someone who actually *knows* in precisely what
> > > ways the Apple "eyeball" calibration method is flawed.
> >
> > I gave you the answer a couple posts ago.
> > Short form: 5 measurement points on the grey axis aren't enough
> > for many screens even *IF* the eye was a perfect instrument.
>
> The human eye is very good at making comparisons and determining when
> two things are the same -- which is the way the Apple method works. Have
> you ever used it?

the human eye is not very good and is very easily fooled.

<http://blogs.discovermagazine.com/badastronomy/2009/06/24/the-blue-and-
the-green/>

<http://www.youtube.com/watch?v=xl1lLze5ZpM>

wait for the animation:
<http://www.klausnordby.com/ego/ill_flash2.html>

squares 'a' and 'b' are the same shade of grey:
<http://i40.tinypic.com/255rtx5.jpg>

> > So if your screen, on the average, is showing hard to tolerate
> > colour shifts and for some colours even completely different
> > colours ... then the calibration method is unusable.
>
> Or (as I suspect might be the case) the screen (and its driving
> hardware) is simply not able to be calibrated properly.

if it can't be calibrated properly, then the display and/or the video
hardware is defective (or the puck).

> Using the Apple calibration method, I can do a perfectly decent job on
> the Dell 2407; all the test images on the web site you referred to look
> just fine. I cannot achieve similar performance from the MacBook's
> display, whether I use the Apple-provided profile, or one of my own --
> though the latter are (I think) somewhat better.

have you measured the delta-e? how do you know how accurate it really
is?

> I have, on several occasions, tried to calibrate other LCD screens as
> well, also mostly with poor results. It is known that LCD screens show
> much more variability in color accuracy than CRTs (usually), and that
> would seem to indicate that not all screens can be properly calibrated.
> Otherwise, why would it even be mentioned in reviews? I bought the 2407
> after learning that it did, indeed, have good color performance.

actually, lcds are *more* accurate than a crt and can have a wider
gamut too. there is also no drift or warm-up period with an lcd as
there is with a crt.

> > > Because from my
> > > understanding of how it works, it should produce pretty decent results.
> >
> > Decent as in 'the grey axis will look quite neutral to the
> > untrained eye, and the colours will be similar', yes.
>
> Well, the Dell 2407 is, and the MacBook is not, capable of producing a
> neutral grey scale; that is precisely the problem. It cannot do that
> even at the five points that you claim are insufficient. I have a
> feeling that if (as you claim) a hardware calibrator can manage to stick
> values in the LUT that the "eyeball" method cannot, the result would be
> better color accuracy at the expense of some other parameter, possibly
> brightness or contrast ratio. Given that choice I (and perhaps Apple)
> would choose a bright display. After all, the Mac is a portable
> computer, not a dedicated photo editor; the screen needs to be usable
> under a wide range of situations.

macbook displays are easily calibrated with a hardware device. they may
not be as good as an eizo, but they work fine.

> > Decent as in 'If you hold your print next to your monitor it'll
> > look the same[1][2]', not in a 1000 years.
>
> FWIW, I'm not interested in hard-copy output; it really is pretty
> limited, both in gamut, and in contrast ratio as well.

then what are you matching?

> > Decent as in 'The monitor will give you a good idea how the print
> > will look like', depends a lot on how good an idea you want.
> >
> > If you've not got a good eye for colour and look at the print
> > a few minutes after you've seen the monitor and don't have any
> > comparisons when looking at the print (i.e. you're not trying to
> > hit a specific colour, say a company logo's colour or a specific
> > flower and thus are not holding the print next to such an object),
> > you will be OK.
> >
> > > Your statements and questions above make it pretty clear that you're not
> > > going to be the one who provides the answer.
> >
> > It seems you're not even getting that you have been handed facts.
> > Or at least that you have problems evaluating what they mean.
> > OR that you are a fanboi and therefore there are no arguments
> > that could make you change your mind, since that would challenge
> > a dogma or something.
>
> If I were a "fanboi", would I be talking about how good the Dell display is,
> and complaining about the one in the Mac?

you keep saying how wonderful the apple method is, but with no proof.

> I just want some facts, specifically explaining *how* the Apple method
> is flawed. All you've done is to say over and over that the mechanized
> method is better. I want to know *why*. Saying that "you get more, and
> more precise, numbers" doesn't mean anything unless you can state
> precisely *why* that difference matters. It's like the audio folks who
> go on about 24 bit samples at 96 or 192 kilosamples/second, and how much
> "better" that is than 16 bit, 44.1 kilosample sound. Sure, the numbers
> are bigger, but in fact, the difference is simply not audible, so who
> cares?

you've been given facts. look at the optical illusions above. your eye
is *easily* fooled. hardware pucks are not.

it's the same with audio. you don't measure frequency or loudness by
ear, you use a frequency counter or sound pressure meter. doctors take
your temperature with a thermometer, not by touching your forehead like
parents sometimes do.

> > If it's the first two, feel free to ask "what does ΔE mean?"
> > or "How do you arrive at these ΔE numbers?" or similar.
>
> I have some knowledge of those, and a few other, things concerning color
> imaging. Prior to retiring, I spent many years in television technology
> including some work with very large screen projection systems, helped
> develop MPEG-2, and designed a few high-performance film printers, among
> other things.

great. what does that have to do with colour management?

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:
> In article <ptvoa9-...@ID-52418.user.berlin.de>,
> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

>> isw <i...@witzend.com> wrote:
>>
>> > I'm still trying to find someone who actually *knows* in precisely what
>> > ways the Apple "eyeball" calibration method is flawed.
>>
>> I gave you the answer a couple posts ago.
>> Short form: 5 measurement points on the grey axis aren't enough
>> for many screens even *IF* the eye was a perfect instrument.

> The human eye is very good at making comparisons and determining when
> two things are the same

Within 1-2 ∆E typical. Consumer sensors are more accurate,
much faster and don't tire.

> -- which is the way the Apple method works.

So how do you compare to, say, a GretagMacbeth Colour Checker
(and that's only 24 patches, not hundreds) in D50 normlight (which
costs more than a simple tristimulus sensor) to your monitor?

So how do you compare the brighness to, say, 120 cd/m²?

How do you comare the chromaticity of the 100% white to the
neutral grey or white in D50 (or D65, or whatever you want to
standardize on)?


All you *can* compare is a relative brightness compared to 0% and
100% (i.e. you have no idea if your screen is too bright!) and
the gray neutrality *compared to* the overly blueish white of
the native monitor setting.

> Have you ever used it?

Does one have to use it *and* not compare the result with an
instrument to have a religious experience?

How about *you* use it and then measure, say, 800 odd patches and
tell us how close *you* came using the eyeball method on *your*
monitor?

>> So if your screen, on the average, is showing hard to tolerate
>> colour shifts and for some colours even completely different
>> colours ... then the calibration method is unusable.

> Or (as I suspect might be the case) the screen (and its driving
> hardware) is simply not able to be calibrated properly.

Yep, that's the "sour grapes" strategy.

Do rent a hardware puck for a couple days and tell us. Your
average photo club, your rental service, etc. all can supply
you with one.

> Using the Apple calibration method, I can do a perfectly decent job on
> the Dell 2407;

So, how high is the average ∆E, how bad the worst ∆E and
please show us an ∆E plot ...

Oh, you don't KNOW!

So how do you *know* you can do a perfectly decent job?

> all the test images on the web site you referred to look
> just fine.

Aha.

BTW, which web site did I refer to?
Oh, but I *didn't*.
So all the imaginary test images on an imaginary website look
'just fine', when you don't even have any idea how they're supposed
to look.

I am sure you didn't have them printed on well calibrated systems
(with enough gamut for the whole image) and compared to your
monitor under normlight.

> I cannot achieve similar performance from the MacBook's
> display, whether I use the Apple-provided profile, or one of my own --
> though the latter are (I think) somewhat better.

You're using very insufficient methods, which may work somewhat
on some displays, but which is not good enough for others.

That's like saying a 2 MPix compact camera doesn't resolve
enough for a 1 meter sized image viewed at close distrance,
THEREFORE it's impossible to create such an image at all.

> I have, on several occasions, tried to calibrate other LCD screens as
> well, also mostly with poor results. It is known that LCD screens show
> much more variability in color accuracy than CRTs (usually),

That's true. How could you have both wide gamut screens and
standard gamut screens otherwise?

> and that
> would seem to indicate that not all screens can be properly calibrated.

Ah ... no. Even systems like printers with 8, 10, 12 colours
can be calibrated --- and that even though they're mixing colours
wildly[1] and they're not exactly known to be monotonous.

So, it's easily true that Apple's eyeball method is not enough
for LCDs (and indeed I said as much), but THAT'S ALL.

> Otherwise, why would it even be mentioned in reviews?

Lots of things are mentioned in reviews.
And some monitors are better (!= good, they're much to bright!) as
they come from the factory than others.

>> > Because from my
>> > understanding of how it works, it should produce pretty decent results.

>> Decent as in 'the grey axis will look quite neutral to the
>> untrained eye, and the colours will be similar', yes.

> Well, the Dell 2407 is, and the MacBook is not, capable of producing a
> neutral grey scale; that is precisely the problem.

The MacBook is perfectly capable of producing a neutral grey
scale.
"After calibration the colour is neutral"
http://forums.whirlpool.net.au/archive/957785

It's your technique that's lacking.

> It cannot do that
> even at the five points that you claim are insufficient.

If you cannot set it to neutral grey because the adjustment
slider area is too small, then it is a fault of your technique.
If you cannot set it to neutral because the adjustment steps are
too large, then it is a fault of your technique.
If you cannot set it to neutral because youc cannot see what
neutral would be, that's again a fault of your technique.

> I have a
> feeling that if (as you claim) a hardware calibrator can manage to stick
> values in the LUT that the "eyeball" method cannot, the result would be
> better color accuracy at the expense of some other parameter, possibly
> brightness or contrast ratio.

That's true: if you need to convert your 9600K monitor white
to D50, you pay with some brightness. If you need to correct
your dark colours, you may need to raise the black point a bit.
Both eat contrast ratio.

Just like a not wide open aperture eats light.

> Given that choice I (and perhaps Apple)
> would choose a bright display.

Then STOP TRYING TO CALIBRATE DISPAYS INCLUDING APPLE'S
"EYEBALL" METHOD!

STOP USING ANY PROFILES!

STOP THIS WHOLE THREAD!

You want overly bright, blue-tinted displays, so that every
print looks too dark and too red? FINE. YOUR PROBLEM.

> After all, the Mac is a portable
> computer, not a dedicated photo editor; the screen needs to be usable
> under a wide range of situations.

Yep, that's why overly bright and overly blue is good. NOT!

Ever heard of being able to remove the settings from the LUT
and reverting to "default values"?

>> Decent as in 'If you hold your print next to your monitor it'll
>> look the same[1][2]', not in a 1000 years.

> FWIW, I'm not interested in hard-copy output; it really is pretty
> limited, both in gamut, and in contrast ratio as well.

Really. That must be because you love overly bright, overly
blue screens.

>> Decent as in 'The monitor will give you a good idea how the print
>> will look like', depends a lot on how good an idea you want.

>> If you've not got a good eye for colour and look at the print
>> a few minutes after you've seen the monitor and don't have any
>> comparisons when looking at the print (i.e. you're not trying to
>> hit a specific colour, say a company logo's colour or a specific
>> flower and thus are not holding the print next to such an object),
>> you will be OK.

>> > Your statements and questions above make it pretty clear that you're not
>> > going to be the one who provides the answer.

>> It seems you're not even getting that you have been handed facts.
>> Or at least that you have problems evaluating what they mean.
>> OR that you are a fanboi and therefore there are no arguments
>> that could make you change your mind, since that would challenge
>> a dogma or something.

> If I were a "fanboi", would I be talking about how good the Dell display is, and
> complaining about the one in the Mac?

You're a fanboi of the Apple eyeball method.

> I just want some facts, specifically explaining *how* the Apple method
> is flawed.

How many repetitions do you want?

> All you've done is to say over and over that the mechanized
> method is better. I want to know *why*.

- the eye isn't an absolute instrument. It cannot tell you if
the monitor is too bright or too blue (or too red) without
something to compare against.
- the eye is less sensitive to colour differences than an
instrument. If you use the measurements to extrapolate to other
colours, you need to be more exact ... the larger the range
of colours you extrapolate to, the more exact you need to be.
There are >16 million colours ...
- 5 measurements are about 50 too few --- and even then you
only have a minimal, rough profile.
- 5 measurements on the grey axis don't say anything about the
other colour axes nor about the rest of the colour space.
- 5 measurements (or 10, as it may be) only allow a gamma or a
shaper profile, and they have assumptions on the display that
said display doesn't necessary fulfill.
- Apple's eyeball method is based on relative brightnesses,
compared to 0% and 100% white. Said white is usually completely
wrong, as it leaves the factory. Connect the dots ...
- Apple's eyeball method isn't good enough for many LCDs, as you
have proven (MacBook screen!)

I guess I said all that a couple times now.

> Saying that "you get more, and
> more precise, numbers" doesn't mean anything unless you can state
> precisely *why* that difference matters.

"A color image which straddles the two screens can be a painful
thing to see ..."

-- isw in Message-ID: <isw-4ECE75.14125826052012@[216.168.3.50]>

It's rather obvious, isn't it?

All you need is 2 proper profiles, one for each monitor ... for
which you need more numbers to remove gamma curve assumptions and
the like and for which you want more precise numbers to have less
extrapolation errors. *Especially* for your laptop you need
many more measurements.

If you were living close enough, I'd come over and calibrate and
profile your screens just to prove my point.

So I can just tell you: rent or borrow a hardware unit and try
for yourself.

> It's like the audio folks who
> go on about 24 bit samples at 96 or 192 kilosamples/second, and how much
> "better" that is than 16 bit, 44.1 kilosample sound. Sure, the numbers
> are bigger, but in fact, the difference is simply not audible, so who
> cares?

Do you know what a brick filter is?
Do you know if it's easier to build if the steepness of the filter
can be reduced?
Do you know why a deep pass filter is needed for digital sampling?

Then you know why more samples/s.

Do you know why some people use RAWs with 12 or 14 or more bits
when JPEGs with visually evenly distributed 8 bit are visually
not different at all (and in fact almost all monitors can
only display 8 bit?)

Then you know why 24 bits sample depth is preferred for some
tasks.

Did you know that sometimes it's much easier to throw more
data into a system than to optimize it to the same result
with less data?

>> If it's the first two, feel free to ask "what does ΔE mean?"
>> or "How do you arrive at these ΔE numbers?" or similar.

> I have some knowledge of those, and a few other, things concerning color
> imaging. Prior to retiring, I spent many years in television technology
> including some work with very large screen projection systems, helped
> develop MPEG-2, and designed a few high-performance film printers, among
> other things.

OK. So it's more a case of obstinacy of old age: you think you
already know everything, so there could be no reason someone else
could realistically offer that would change your mind. I still
hope it's not the reason, but as they say, hope's the last thing
to die.

Please report back with the ∆E values of your 'calibrated'
screen.

Or at least tell us with your knowledge which 5 points I
should pick for calibrating my LCD monitor ...

-Wolfgang

[1] for example they're mixing additional CMY (and their light
versions) to the black (and grey and light grey) inks to
reduce speckle (just one or two black or grey dots look more
irregular than a whole handful of dithered LM, LC and Y dots)
and black is even blacker if CMY is mixed into it as well.

Oh, and of course there's no perfectly neutral grey, so there
CMY (or their light versions) needs to be mixed in, too.

[isw]

In article <83lsa9-...@ID-52418.user.berlin.de>,

Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:

> isw <i...@witzend.com> wrote:
> > In article <ptvoa9-...@ID-52418.user.berlin.de>,
> > Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>
> >> isw <i...@witzend.com> wrote:
> >>
> >> > I'm still trying to find someone who actually *knows* in precisely what
> >> > ways the Apple "eyeball" calibration method is flawed.

--snip--

> > Have you ever used it?
>
> Does one have to use it *and* not compare the result with an
> instrument to have a religious experience?

Well, it's difficult to understand how you can know how bad it is if you
don't even know *what* it is or how it works ...

-- snip --

> >> Decent as in 'If you hold your print next to your monitor it'll
> >> look the same[1][2]', not in a 1000 years.
>
> > FWIW, I'm not interested in hard-copy output; it really is pretty
> > limited, both in gamut, and in contrast ratio as well.
>
> Really. That must be because you love overly bright, overly
> blue screens.

Actually, it's because I know how poor color reproduction is, using CMYK
inks.

-- snip --

> > It's like the audio folks who
> > go on about 24 bit samples at 96 or 192 kilosamples/second, and how much
> > "better" that is than 16 bit, 44.1 kilosample sound. Sure, the numbers
> > are bigger, but in fact, the difference is simply not audible, so who
> > cares?
>
> Do you know what a brick filter is?
> Do you know if it's easier to build if the steepness of the filter
> can be reduced?
> Do you know why a deep pass filter is needed for digital sampling?

I was designing sharp-cutoff phase-corrected analog filters for
broadcast applications over forty years ago. I have designed both anti-
aliasing filters for encoding, and reconstruction filters for decoding,
digital signals. Not to mention antialiasing of computer-created
photographic images by the proper overlap of the gaussian brightness
distribution of the CRT spot. What's your point?

> Then you know why more samples/s.

I know that once you have exceeded the resolution capabilities of the
final "receiver" (in this case human perception), either in amplitude
precision or time precision, more bits in either dimension are
completely pointless.

> Do you know why some people use RAWs with 12 or 14 or more bits
> when JPEGs with visually evenly distributed 8 bit are visually
> not different at all (and in fact almost all monitors can
> only display 8 bit?)
>
> Then you know why 24 bits sample depth is preferred for some
> tasks.

And is totally unnecessary for final viewing or listening.

> Did you know that sometimes it's much easier to throw more
> data into a system than to optimize it to the same result
> with less data?

Easier, and a method commonly used by those who don't understand things
very well, but usually not better, for lots of reasons.

Isaac

[isw]

In article <150620121203312995%nos...@nospam.invalid>,

nospam <nos...@nospam.invalid> wrote:

> In article <isw-0B81EF.00344215062012@[216.168.3.50]>, isw
> <i...@witzend.com> wrote:
>
> > > > I'm still trying to find someone who actually *knows* in precisely what
> > > > ways the Apple "eyeball" calibration method is flawed.

-- snip --

> actually, lcds are *more* accurate than a crt and can have a wider
> gamut too. there is also no drift or warm-up period with an lcd as
> there is with a crt.

That is not consistent with a whole lot of information I read when I was
trying to decide whether to keep my trusty CRT going, or replace it with
a flat panel. My own personal experience with fluorescent light sources
is that they do indeed drift, both in intensity and spectrally, over
time.

-- snip --

> then what are you matching?

Two side-by-side LCDs -- the Dell 2407 and the one in the MacBook.

-- snip --

> you keep saying how wonderful the apple method is, but with no proof.

I've never said it was "wonderful". I've been told that it was "no
good", and I'm just trying to find out *why*. I have described the
results to some folks who, best I can figure out, have never tried it,
and know nothing about it, not even the theory behind it. Other than how
bad it is ...

-- snip --

> it's the same with audio. you don't measure frequency or loudness by
> ear, you use a frequency counter or sound pressure meter.

The point is that in both cases, measuring to arbitrarily great
precision is possible, but the question (unanswered) is *at what
precision is human visual resolution equalled or exceeded*? Because that
is a limit that it's simply not necessary to exceed.

I'm perfectly aware that a mechanical calibrator can deliver more
"accuracy". What nobody who says it is bad seems to know is *in what
ways* the Apple method is deficient.

Isaac

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:
> nospam <nos...@nospam.invalid> wrote:

>> isw <i...@witzend.com> wrote:

> -- snip --

>> you keep saying how wonderful the apple method is, but with no proof.

> I've never said it was "wonderful". I've been told that it was "no
> good", and I'm just trying to find out *why*. I have described the
> results to some folks who, best I can figure out, have never tried it,
> and know nothing about it, not even the theory behind it. Other than how
> bad it is ...

Well, I know the theory, I know what it matches, and I have
*proved* that even given very accurate measurements (way better
than the eye produces) it's not very good. Between "just
acceptable" and "completely unusable" for colour matching.

All you've been doing is handwaving that another method which
you have never tried, don't understand what it's doing and
know not a shred of theory about cannot be visually better,
won't work on your laptop screen anyways and is wholly
unneccessary.

I've provided data.
You've provided more handwaving.


So is that a case of the pot calling the bridal gown black?

> -- snip --

>> it's the same with audio. you don't measure frequency or loudness by
>> ear, you use a frequency counter or sound pressure meter.

> The point is that in both cases, measuring to arbitrarily great
> precision is possible, but the question (unanswered) is *at what
> precision is human visual resolution equalled or exceeded*?

That has been answered. Roughly around 1 ∆E, depending on the
specific ∆E used and the colour and brightness. Later ∆E
definitions are better in that regard, like the 2000'er definition.

> Because that
> is a limit that it's simply not necessary to exceed.

*sigh*
If you're extrapolating --- and you are, since you're not checking
16 million colours --- any error is multiplied with the distance.

So, yes, if you were checking every colour, you'd need not
to exceed 1 ∆E. Since with the Apple eyeball method you're
extrapolating from 50% gray to 89% blue, you need way more accurate
measurements or way closer measurements. An instrument checking
hundreds of patches provides both.

> I'm perfectly aware that a mechanical calibrator can deliver more
> "accuracy". What nobody who says it is bad seems to know is *in what
> ways* the Apple method is deficient.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.


I start to feel like a broken record.
Either you're not even grasping the slightest bit of what I am
saying, so you cannot even ask intelligent questions, or you're
trolling. Maybe English isn't your first language?

-Wolfgang

[Wolfgang Weisselberg]

nospam <nos...@nospam.invalid> wrote:
> isw <i...@witzend.com> wrote:

>> I have, on several occasions, tried to calibrate other LCD screens as
>> well, also mostly with poor results. It is known that LCD screens show
>> much more variability in color accuracy than CRTs (usually), and that
>> would seem to indicate that not all screens can be properly calibrated.
>> Otherwise, why would it even be mentioned in reviews? I bought the 2407
>> after learning that it did, indeed, have good color performance.

> actually, lcds are *more* accurate than a crt

By what measure?
They're often more well behaved ... but that's all.

> there is also no drift or warm-up period with an lcd as
> there is with a crt.

CCFL does need a 30 minute warm up period to become stable.
Additionally, CRTs usually have identical brightness all over
the screen, this is not the case for most LCDs ...

> it's the same with audio. you don't measure frequency or loudness by
> ear, you use a frequency counter or sound pressure meter. doctors take
> your temperature with a thermometer, not by touching your forehead like
> parents sometimes do.

Oh, touching the forehead will do as a rough check if the
person is *way* out of the normal temperature range, given
good enough circumstances.

-Wolfgang

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:
> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>> isw <i...@witzend.com> wrote:
>> > Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>> >> isw <i...@witzend.com> wrote:

>> >> > I'm still trying to find someone who actually *knows* in precisely what
>> >> > ways the Apple "eyeball" calibration method is flawed.

> --snip--

>> > Have you ever used it?

>> Does one have to use it *and* not compare the result with an
>> instrument to have a religious experience?

> Well, it's difficult to understand how you can know how bad it is if you
> don't even know *what* it is or how it works ...

Well, I know *what* a Saturn V is and how it works. Yet I've
never ridden a rocket ... can you spot the error in your logic?

If anything I have done a better test than what Apple's
eyeball offers, with twice as many measurements and more
accurate data ...

> -- snip --

>> >> Decent as in 'If you hold your print next to your monitor it'll
>> >> look the same[1][2]', not in a 1000 years.

>> > FWIW, I'm not interested in hard-copy output; it really is pretty
>> > limited, both in gamut, and in contrast ratio as well.

>> Really. That must be because you love overly bright, overly
>> blue screens.

> Actually, it's because I know how poor color reproduction is, using CMYK
> inks.

Try CcMmYRKLKLLK. Or CcMmYKRG.
Or use a RGB process --- like photosensitive paper, chemically
developed, just like in the days of yore.

And even basic CMYK doesn't automatically mean poor color
reproduction, just a somewhat more limited gamut. (And yet it'll
display colours that are not available for your average screen.)

> -- snip --

>> > It's like the audio folks who
>> > go on about 24 bit samples at 96 or 192 kilosamples/second, and how much
>> > "better" that is than 16 bit, 44.1 kilosample sound. Sure, the numbers
>> > are bigger, but in fact, the difference is simply not audible, so who
>> > cares?

>> Do you know what a brick filter is?
>> Do you know if it's easier to build if the steepness of the filter
>> can be reduced?
>> Do you know why a deep pass filter is needed for digital sampling?

> I was designing sharp-cutoff phase-corrected analog filters for
> broadcast applications over forty years ago.

How sharp, how much supression, at what price?

> I have designed both anti-
> aliasing filters for encoding, and reconstruction filters for decoding,
> digital signals. Not to mention antialiasing of computer-created
> photographic images by the proper overlap of the gaussian brightness
> distribution of the CRT spot. What's your point?

OK, then you should know why oversampling is a good idea.

>> Then you know why more samples/s.

> I know that once you have exceeded the resolution capabilities of the
> final "receiver" (in this case human perception), either in amplitude
> precision or time precision, more bits in either dimension are
> completely pointless.

So your average stereo system does not come with eqalizers?
Never mind the high end systems ...

>> Do you know why some people use RAWs with 12 or 14 or more bits
>> when JPEGs with visually evenly distributed 8 bit are visually
>> not different at all (and in fact almost all monitors can
>> only display 8 bit?)

>> Then you know why 24 bits sample depth is preferred for some
>> tasks.

> And is totally unnecessary for final viewing or listening.

Only when you're using a format that needs absolutely no more
shaping or tweaking (say, colour management or equalizers).
The more you need to shape your signal, the stronger some
parts will be stretched, soon causing visible (or audible)
steps between minimal increments.

Audio needs much less information than images, so it's much easier
to just throw in a couple more bits and to use systems that are
cheaper, but need more signal shaping. Images just need too much
space for that to work easily.

>> Did you know that sometimes it's much easier to throw more
>> data into a system than to optimize it to the same result
>> with less data?

> Easier, and a method commonly used by those who don't understand things
> very well, but usually not better, for lots of reasons.

At least that's your story, and your indignity at people going the
'easy' way and arriving with better results than you do.

-Wolfgang

[Eric Stevens]

On Sun, 17 Jun 2012 13:17:31 +0200, Wolfgang Weisselberg
<ozcv...@sneakemail.com> wrote:

>isw <i...@witzend.com> wrote:
>> nospam <nos...@nospam.invalid> wrote:
>>> isw <i...@witzend.com> wrote:
>
>> -- snip --
>
>>> you keep saying how wonderful the apple method is, but with no proof.
>
>> I've never said it was "wonderful". I've been told that it was "no
>> good", and I'm just trying to find out *why*. I have described the
>> results to some folks who, best I can figure out, have never tried it,
>> and know nothing about it, not even the theory behind it. Other than how
>> bad it is ...
>
>Well, I know the theory, I know what it matches, and I have
>*proved* that even given very accurate measurements (way better
>than the eye produces) it's not very good. Between "just
>acceptable" and "completely unusable" for colour matching.
>
>All you've been doing is handwaving that another method which
>you have never tried, don't understand what it's doing and
>know not a shred of theory about cannot be visually better,
>won't work on your laptop screen anyways and is wholly
>unneccessary.

The problem is that most of us are writing about methods designed as
the first stage of a properly colour-managed work flow. We are trying
to ensure that the colours seen on our screen are as near as possible
to the colours which will be seen on another correctly
colour-calibrated screen and as close as possible emerging from a
correctly colour-calibrated print process.

This is not what 'isw' (who seems to have taken over from the OP) is
after. All he wants is a screen which he can calibrate ('just fine')
to look 'good'. The fact that it might not look the same as any other
screen does not seem to bother him, although it would bother most of
the rest of us.

We will never agree while we are talking at cross purposes.

>
>I've provided data.
>You've provided more handwaving.
>
>
>So is that a case of the pot calling the bridal gown black?
>
>
>> -- snip --
>
>>> it's the same with audio. you don't measure frequency or loudness by
>>> ear, you use a frequency counter or sound pressure meter.
>
>> The point is that in both cases, measuring to arbitrarily great
>> precision is possible, but the question (unanswered) is *at what
>> precision is human visual resolution equalled or exceeded*?
>

>That has been answered. Roughly around 1 ?E, depending on the
>specific ?E used and the colour and brightness. Later ?E

>definitions are better in that regard, like the 2000'er definition.
>
>> Because that
>> is a limit that it's simply not necessary to exceed.
>
>*sigh*
>If you're extrapolating --- and you are, since you're not checking
>16 million colours --- any error is multiplied with the distance.
>
>So, yes, if you were checking every colour, you'd need not

>to exceed 1 ?E. Since with the Apple eyeball method you're

>extrapolating from 50% gray to 89% blue, you need way more accurate
>measurements or way closer measurements. An instrument checking
>hundreds of patches provides both.
>
>> I'm perfectly aware that a mechanical calibrator can deliver more
>> "accuracy". What nobody who says it is bad seems to know is *in what
>> ways* the Apple method is deficient.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>See all of my other posts.
>
>
>I start to feel like a broken record.
>Either you're not even grasping the slightest bit of what I am
>saying, so you cannot even ask intelligent questions, or you're
>trolling. Maybe English isn't your first language?
>
>-Wolfgang

Regards,

Eric Stevens

[nospam]

In article <isw-07399C.23355515062012@[216.168.3.50]>, isw

<i...@witzend.com> wrote:

> > you keep saying how wonderful the apple method is, but with no proof.
>
> I've never said it was "wonderful". I've been told that it was "no
> good", and I'm just trying to find out *why*. I have described the
> results to some folks who, best I can figure out, have never tried it,
> and know nothing about it, not even the theory behind it. Other than how
> bad it is ...

it's not that it's 'no good' it's that it isn't *as* good as a hardware
puck. this should be obvious, without any of the proof that has been
provided.

have you tried a hardware puck? if you have, you'd see the difference
and this discussion would not be taking place. whether you care about
the difference is another story.

[isw]

In article <170620122204568836%nos...@nospam.invalid>,

nospam <nos...@nospam.invalid> wrote:

> In article <isw-07399C.23355515062012@[216.168.3.50]>, isw
> <i...@witzend.com> wrote:
>
> > > you keep saying how wonderful the apple method is, but with no proof.
> >
> > I've never said it was "wonderful". I've been told that it was "no
> > good", and I'm just trying to find out *why*. I have described the
> > results to some folks who, best I can figure out, have never tried it,
> > and know nothing about it, not even the theory behind it. Other than how
> > bad it is ...
>
> it's not that it's 'no good' it's that it isn't *as* good as a hardware
> puck. this should be obvious, without any of the proof that has been
> provided.

More precision is in some way "better". The question I still have is,
*how* is the Apple eyeball method deficient -- what errors will it make,
that a machine will not?

> have you tried a hardware puck? if you have, you'd see the difference
> and this discussion would not be taking place. whether you care about
> the difference is another story.

I want to find out if the investment is worth the money before I buy
one. To date, nobody has been able to tell me, chapter and verse,
precisely what is wrong with the Apple method. I'm beginning to suspect
that nobody knows ...

Isaac

[isw]

In article <bvq0b9-...@ID-52418.user.berlin.de>,

Nobody I know can differentiate 16 million colors; I doubt that yo know
any such person either. That goes to my point about the nonsense of
"more digits".

> So, yes, if you were checking every colour, you'd need not
> to exceed 1 ∆E. Since with the Apple eyeball method you're
> extrapolating from 50% gray to 89% blue, you need way more accurate
> measurements or way closer measurements. An instrument checking
> hundreds of patches provides both.
>
> I'm perfectly aware that a mechanical calibrator can deliver more
> "accuracy". What nobody who says it is bad seems to know is *in what
> ways* the Apple method is deficient.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.

See all of my other posts.


I start to feel like a broken record.

Here's an example of the sort of thing thing you have not answered: What
are the expected errors of the Apple method, compared to the errors of a
mechanical calibrator? IOW, *how bad is it*?

> Either you're not even grasping the slightest bit of what I am
> saying, so you cannot even ask intelligent questions, or you're
> trolling. Maybe English isn't your first language?

If anything, "science" is my native language. Evidence works very well
for me. I still do not have any evidence showing in which ways the Apple
method is deficient (other than "less precision", which by itself is
meaningless). I'm beginning to think you don't know either ...

Isaac

[nospam]

In article <isw-AF24F1.21404517062012@[216.168.3.50]>, isw

<i...@witzend.com> wrote:

> > > > you keep saying how wonderful the apple method is, but with no proof.
> > >
> > > I've never said it was "wonderful". I've been told that it was "no
> > > good", and I'm just trying to find out *why*. I have described the
> > > results to some folks who, best I can figure out, have never tried it,
> > > and know nothing about it, not even the theory behind it. Other than how
> > > bad it is ...
> >
> > it's not that it's 'no good' it's that it isn't *as* good as a hardware
> > puck. this should be obvious, without any of the proof that has been
> > provided.
>
> More precision is in some way "better". The question I still have is,
> *how* is the Apple eyeball method deficient -- what errors will it make,
> that a machine will not?

because your eye is very easily fooled. look at the optical illusions i
posted.

> > have you tried a hardware puck? if you have, you'd see the difference
> > and this discussion would not be taking place. whether you care about
> > the difference is another story.
>
> I want to find out if the investment is worth the money before I buy
> one. To date, nobody has been able to tell me, chapter and verse,
> precisely what is wrong with the Apple method. I'm beginning to suspect
> that nobody knows ...

people have told you, you just don't want to hear it.

[Savageduck]

On 2012-06-17 21:50:15 -0700, isw <i...@witzend.com> said:


>
> I start to feel like a broken record.
>
> Here's an example of the sort of thing thing you have not answered: What
> are the expected errors of the Apple method, compared to the errors of a
> mechanical calibrator? IOW, *how bad is it*?

The Apple method is not bad at all. The only problem is it is dependent
on the availability of a healthy, well calibrated MkI eyeball.
For the majority of users with young heathy vision they are probably
able to calibrate their display perfectly well using the Apple method.
That said there is a sector of the general population with less than
perfect, well calibrated MkI eye balls.
I have used the Apple method for years with what I believed were
satisfactory results. I was wrong, my various vision issues, myopia and
quite extreme astigmatism have for years been corrected with
progressive lenses. I have learned that my ability to accurately
calibrate using the Apple method with my out of whack eye balls and
their extreme corrective lenses was compromised. So I stepped up and
bought a Pantone huey Pro, which has made life simple. I update the
profile every two weeks and the huey sensor and software constantly
adjust for changing light. The result is I have a Mac which maintains a
constant calibrated display under all lighting conditions. I have had
no questionable discrepancies between screen & print for some time now.
I can get to work with PS without thinking, "Should I run a quick eye
ball display calibration before I get started.?"
I can believe that what I see is what I get.

Having said all that, I would imagine that the majority will do just
fine using the Apple method.

>> Either you're not even grasping the slightest bit of what I am
>> saying, so you cannot even ask intelligent questions, or you're
>> trolling. Maybe English isn't your first language?
>
> If anything, "science" is my native language. Evidence works very well
> for me. I still do not have any evidence showing in which ways the Apple
> method is deficient (other than "less precision", which by itself is
> meaningless). I'm beginning to think you don't know either ...
>
> Isaac

Trust in your eyeball is the key to the Apple method, and for those
with questionable vision issues, a mechanical calibration is probably a
wise decision.


--
Regards,

Savageduck

[Eric Stevens]

Of course they don't and I suspect that you know perfectly well why.

The ability of the Apple method to colour-calibrate a screen depends
upon the quality of what Duck refers to as the 'Mk1 eyeball'. These
vary enormously from person to person and there is no single numerical
measure of the ability of the individual eyeball to function as an
element in the Apple method.

The question you keep asking is meaningless and that's why you will
never get a satisfactory answer to it. The best that you can hope for
is a statement that as far as a particular individual concerned there
is no way of predicting the accuracy of the Apple method.

Regards,

Eric Stevens

[Rob]

Just to reiterate on the MK1 Eye Ball, that there was a famous artist
who over time was introducing more (I think) Magenta to his works.

This showed how his perception of colour was changing over time.

[Eric Stevens]

You may be thinking of Claude Monet
http://www.psych.ucalgary.ca/pace/va-lab/avde-website/monet.html
A 'graphic' illustration of the futility of relying on the eye for
colour calibration.

Regards,

Eric Stevens

[Frank S]

"Eric Stevens" <eric.s...@sum.co.nz> wrote in message
news:f2o1u7polu76tm676...@4ax.com...

I remember reading there was a possibility van Gogh's paintings showed an
increase in yellows because of his medication by digitalis in later years.

--
Frank ess

[Chris Malcolm]

How? Because no matter how his colour perception changed, it would
change the colour of his painting and the colour of the subject of the
painting in the same way. So just as with looking through rose tinted
glasses, he'd have to make the painting the same colour as the subject
in order for it to look the same to him. So it would also look the
same colour to normal vision.

--
Chris Malcolm

[Wolfgang Weisselberg]

isw <i...@witzend.com> wrote:
> Wolfgang Weisselberg <ozcv...@sneakemail.com> wrote:
>> isw <i...@witzend.com> wrote:
>> > nospam <nos...@nospam.invalid> wrote:
>> >> isw <i...@witzend.com> wrote:

[snip]

>> I've provided data.
>> You've provided more handwaving.

>> So is that a case of the pot calling the bridal gown black?

No answer ...
so how about some data showing your point of view?
Aren't you a sciency-evidency guy?

>> That has been answered. Roughly around 1 ∆E, depending on the
>> specific ∆E used and the colour and brightness. Later ∆E
>> definitions are better in that regard, like the 2000'er definition.

>> > Because that
>> > is a limit that it's simply not necessary to exceed.

>> *sigh*
>> If you're extrapolating --- and you are, since you're not checking
>> 16 million colours --- any error is multiplied with the distance.

> Nobody I know can differentiate 16 million colors;

Irrelevant. Even if you can see only 10 million (current claims)
or 1 million or 100,000 colours, you're still extrapolating to
the *same gamut* and the *same distance* and you still only have
5 measurement points from which you extrapolate.

And the computer *has* to extrapolate to 2^8^3 colours ...

>> So, yes, if you were checking every colour, you'd need not
>> to exceed 1 ∆E. Since with the Apple eyeball method you're
>> extrapolating from 50% gray to 89% blue, you need way more accurate
>> measurements or way closer measurements. An instrument checking
>> hundreds of patches provides both.

No answer. Hmmm ...


[snip]

> Here's an example of the sort of thing thing you have not answered: What
> are the expected errors of the Apple method, compared to the errors of a
> mechanical calibrator? IOW, *how bad is it*?

For your MacBook you can see for yourself how bad it is.

For a CRT and a Laptop screen I already gave you the best case
∆E for the method, for the latter the whole dataset.

I already gave you the meaning of the different ∆E values.

Maybe you need "average: well below 1 ∆E, in the sub-0.5∆E
region, with the maximum still well controlled below 2-3∆E"
as the answer for "how good is the hardware method".

>> Either you're not even grasping the slightest bit of what I am
>> saying, so you cannot even ask intelligent questions, or you're
>> trolling. Maybe English isn't your first language?

> If anything, "science" is my native language.

Sure, and I'm writing Kiswahili here.
If you were 'science', you'd be into measuring instead of
eyeballing your screen.

> Evidence works very well for me.

Evidence shows that evidence doesn't work for you at all.

Proof: You claim your MacBook screen cannot be calibrated, when
evidence shows it can (it just isn't a very good screen).
Proof: You got all the numbers (and the visualisation) for one
screen and don't do anything with it, you just ignore it.
Proof: Even after being told that you should rent a hardware
device and see for yourself --- you ignore it.

If you were into evidence you'd be able to say what sort of
evidence you'd accept, or say in detail why you don't accept
the facts presented by e.g. me.

> I still do not have any evidence showing in which ways the Apple
> method is deficient (other than "less precision", which by itself is
> meaningless).

You have all the data to evaluate what 'less precision'
implies. You have ∆E numbers and what they mean.
You have a striking personal example what "less precision" means
--- just split a colour image over your desktop and laptop screen.

You have another striking example that 'less precision' in case
of your laptop screen means, you yourself think that it's unable
to be calibrated ... when it can be calibrated. (Just ask Google!)

> I'm beginning to think you don't know either ...

I gave you everything you need to comprehend.

I've almost finished the thought that you're either a troll or
incapable of seeing and understanding evidence.

-Wolfgang

[Wolfgang Weisselberg]

Eric Stevens <eric.s...@sum.co.nz> wrote:
> On Sun, 17 Jun 2012 21:40:45 -0700, isw <i...@witzend.com> wrote:
>> nospam <nos...@nospam.invalid> wrote:

>>> have you tried a hardware puck? if you have, you'd see the difference
>>> and this discussion would not be taking place. whether you care about
>>> the difference is another story.

>>I want to find out if the investment is worth the money before I buy
>>one.

Rent one.

[snip]

> The ability of the Apple method to colour-calibrate a screen depends
> upon the quality of what Duck refers to as the 'Mk1 eyeball'.

Having tried the method --- with a hardware calibrator instead
of an eye! --- it's not just the eyeball that's in question.
The monitor must be well behaved for Apple's method to work.

I did publish one set of data here ... see for yourself.

-Wolfgang

[Wolfgang Weisselberg]

Savageduck <savageduck1@{REMOVESPAM}me.com> wrote:
> On 2012-06-17 21:50:15 -0700, isw <i...@witzend.com> said:

>>> I start to feel like a broken record.

>> Here's an example of the sort of thing thing you have not answered: What
>> are the expected errors of the Apple method, compared to the errors of a
>> mechanical calibrator? IOW, *how bad is it*?

> The Apple method is not bad at all.

It's much better than nothing at all, yes. As isw writes, it
fails completely for his MacBook.

> The only problem is it is dependent
> on the availability of a healthy, well calibrated MkI eyeball.

And on a very well behaved monitor.
Apparently the MacBook doesn't have such a beast.

> For the majority of users with young heathy vision they are probably
> able to calibrate their display perfectly well using the Apple method.

Depends on your definition of "perfectly well".
If you're trying to hit the right skin colour in mixed light[1]
you do depend on your MkI eyeball ... *and* on the correct display
of that non-gray skin colours on your monitor.

It's better not to be much hindered by a none-too-perfect monitor.

> calibrate using the Apple method with my out of whack eye balls and
> their extreme corrective lenses was compromised. So I stepped up and
> bought a Pantone huey Pro, which has made life simple. I update the
> profile every two weeks and the huey sensor and software constantly
> adjust for changing light.

Ah, if
http://www.behardware.com/articles/626-7/monitor-calibration-for-0-to-100.html
is correct (and if you're using the same software), the software is
... well, suboptimal. (But then you can always use argyllcms.com
for free.) Way too few patches nor a proper choice of patches for
the calibration method.

> The result is I have a Mac which maintains a
> constant calibrated display under all lighting conditions. I have had
> no questionable discrepancies between screen & print for some time now.
> I can get to work with PS without thinking, "Should I run a quick eye
> ball display calibration before I get started.?"
> I can believe that what I see is what I get.

Well, maybe you lucked out with your monitor. :-)

> Having said all that, I would imagine that the majority will do just
> fine using the Apple method.

Yep, but isw's MacBook ...

> Trust in your eyeball

... and trust in your monitor ...

> is the key to the Apple method, and for those
> with questionable vision issues,

... or less perfect monitors ...

> a mechanical calibration is probably a
> wise decision.

And it's so much less straining to just let the computer run the
tests ...

-Wolfgang

[1] and you don't have a grey card (or, even better, a
colour card) shot at the same mixed light situation to
correct for at least the light from one direction ...
say because the athlete or politican or musicians are
running or making a speech or playing and you wouldn't
make friends by intruding ...