Web-based monitor calibration?

[Bob Hill]

Hi Jon,
Not sure this helps, but Eizo sells a line of monitors with automated color calibration using a device that swings down out of the housing of the monitor. The calibration is independent of a computer platform and can run cal checks by the number of hours or on a calendar basis. I realize this is not a solution for those who have monitors in place, buit it may be one going forward.

I saw this monitor at the Color Management Conference, however, I'm no display expert so the value of that feature would have to be weighed against it's ability to properly calibrate the display. I'll leave that question to folks like you.

[Jon Meyer]

Introduction:
Monitor calibration is based on the following:
1) Building unique look up table for the video card - Assists in creating linear gray balance as well as targeting colors within the monitor's gamut
2) Brightness Normalization - Specific to a room's brightness and image representation (brighter for brightly lit rooms, brighter for web purposes, darker for print simulation or darker for a dimly lit room)
3) Contrast - Usually 2.2 Gamma or L* for those who critically judge contrast steps
4) ICC Profile - When used in combination with the above 3 calibrations, will produce the best viewing result - IF - the image software calls the ICC profile into use. (Many internet browsers do not. Photoshop and other pieces of the Adobe Creative Suite do.)

Calibration and profiling can use either a spectrophotometer or a colorimeter.

A spectro sometimes reads too many samples and then the software errors in its averaging algorithms. A less expensive colorimeter will typically produce equal or sometimes even better results. (more info on request)

Several high end displays have built in colorimeters and will calibrate remotely upon an administrator's instruction. This is also true of the medical field where doctors view results - but typically do not perform their hardware calibration.

Using that concept, inexpensive colorimeters could be purchased and periodically placed over a monitor for a Java based web app to communicate with.

At the end of the day, who cares if it is web based, or local - considering that the real cost is in the colorimeter?

Summary: Real monitor calibration and profile building requires a device more sensitive than human vision. The human eye is too adaptive to change for vision to be part of a critical color calibration system.

[MAC.NETWORK]

I was also checking the dates on this post - dangerously close to April 1st. zetaprints is of course correct, Al Ferrari just isn't thinking straight - no adjustment of the monitor controls will affect the screen grab. Turning the monitor off will still give the same screen grab! But Apple's OSX monitor calibration could work in a similar way using a web site.
OSX asks the user to evaluate several test images and uses this to set up the monitor so that shadows are not too dark and and all steps in a grey scale are discernible. The Gamma is adjusted with a full range photo and colour casts evaluated by comparison.

It's a rough and ready method and I would guess something similar could be transferred to a web site.

[KamilT]

Hi all,

I remembered old debate in another forum where I mentioned the surprisingly good result of Supercal program. And yes, it still says the result is comparable to using some "less good" devices (I will not mention, of course. So see for yourself, I think it's a pretty good program ...

http://www.bergdesign.com/supercal/

P.S: Mac only, sorry

Best regards
Kamil Tresnak, Prague

[meddington]

Without a measurement of the emissive output of the screen, this seems like an exercise in futility. If accuracy is even remotely important, even MacGyver would spring for an i1 display.

[KamilT]

Good point Meddington
OK, i1 is not on my list of "hated" devices

But just for sure, did you tried this sw? ... trust me, you will be surprised. We are not discussing if measuring is importatnt, this is without debate, but original post was in "just curious ..." style.
Sure, there are several disadvantages, but trust me, you will be suprised with result.

Best regards
K.

Without a measurement of the emissive output of the screen, this seems like an exercise in futility. If accuracy is even remotely important, even MacGyver would spring for an i1 display.

[guncis]

Hi, everyone

Keep it simple, cause otherwise we will come to real monitor calibration.
Claim - you do not give any warranty on such calibration method
------------ simple idea
Just put on screen some grey and CMYK patches with descriptions
There MUST be -
Grey patches in at least 25,50,75% side by side (or similar)
light and dark CMYK patches (with text - you have to see them and to see difference)
Some human photos
Some different well known colors - like Milka, Coca-cola etc
Some error situations (wrong grey balance, cannot see 95% of Cyan etc) and instructions how to adjust monitor if something is wrong (some small videos would be fine)
All this you can put in one screen
!!! make sure show this in RGB and make on-screen proof in RGB, otherwise diffferent settings can cause different results (you have to convert CMYK->RGB before post on screen proof otherwise you need Serendipity, Dalim or similar s/w)
---------------------
When customer makes all this according to instructions and acceptable for his eyes - monitor WILL NOT BE CALIBRATED, but atleast they will not miss important things.

** just my some unprofessional thoughts

[guncis]

Forget one more thing - tell about right profiles in right places. This helps a lot as well
You can split instructions by "dummy" and "wizard". My thoughts is something between them.

[michaelejahn]

OK, those are good crits. I like them. But let's imagine for a moment that we had a screen capture app that did not introduce a shift. Isn't Photoshop the proper non shifting display companion app for this exercise?

So if we had the ideal shift free capture app and a display app with the capability to measure, such as Photoshop, then why would round tripping not be possible???

Photoshop is dependent on your color settings to display values, yes ? Can you work in Black and White properly ? I would believe in the Easter Bunny and the tooth fairy far before I would believe Photoshop for 'numbers' used for calibration !

Are you familiar with Lab mode in Photoshop ?- i think we need to start there to help you understand a few basic problems...

from Iliah Borg...

B/w image in Lab
http://pics.livejournal.com/sail2ithaki/pic/0003rgzp

Just L channel made visible, brightness/contrast are wrong
http://pics.livejournal.com/sail2ithaki/pic/0003shgf

Same, but with Lstar profile set as the grey working space
http://pics.livejournal.com/sail2ithaki/pic/0003xd7w

Same, but with Color Preferences closed, brightness/contrast are right
http://pics.livejournal.com/sail2ithaki/pic/0003y2rg

Dropping to ridiculously low gamma grey, all is wrong
http://pics.livejournal.com/sail2ithaki/pic/0003wpqf

Switching on Show Color Channels in Color
http://pics.livejournal.com/sail2ithaki/pic/00041sfs

L channel with grey profile set to ridiculous and Show Channels in Color switched on, brightness/contrast are right.
http://pics.livejournal.com/sail2ithaki/pic/00043428

Here is a comment (about the above) from a the Color Theory group (by Jacob Rus)

it’s quite unfortunate that (at least as far as I know) there’s
no way to get Photoshop’s gray working space to be similar to L*. [For
me, the “other” item in the selector menu for gray working space in
the color settings is grayed out.] The "Gray gamma 2.2" space is
somewhat lighter but not entirely dissimilar, and "Dot Gain 30%" has
similar average lightness, but has more contrast in the midtones and
correspondingly less in the highs and lows (like an "s-shaped curve").
I set my working spaces up as sRGB, GRACoL, Gray Gamma 2.2, and then
just mostly work in CIELAB space.

It seems pretty silly to me that viewing a single channel at once as
grayscale always shows it as if the content was in the current gray
working space, when in CIELAB mode the obvious "right thing" to do is
show L* as L*. Alas. I guess they want to masks &c. in the same space,
and for such it's not quite as clear to me what the best choice is.
Display as Gray Gamma 2.2 works just fine for me there; any choice of
gray working space would probably be fine for showing
masks/selections/etc.

[graficworx]

Zeta is right, the RGB screen capture will be the image your video card sees, not what the monitor displays.