Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: ato on October 09, 2006, 10:28:16 pm
-
i've test 2 software with my eye-one display 2,basiccolor 4.0.2 and i1 match3.6.1.
after hardware calibration both software create a profile and load in my display card.
i test the delta E between load/unload the LUT
loaded
(http://www.atoyu.com/image/load_lut.jpg)
unload
(http://www.atoyu.com/image/unload_lut.jpg)
you can see the value get high,so does it mean if i using hardware calibration still need a display card do a correction first to have better result??
if not,why they create a profile and load in to my display card?
-
you can see the value get high,so does it mean if i using hardware calibration still need a display card do a correction first to have better result??
[a href=\"index.php?act=findpost&pid=79746\"][{POST_SNAPBACK}][/a]
These validation results merely show how well the created profile models the actual displayed results. They say nothing about the accuracy of the calibration ... though, in general, the more linear your monitor the lower the deviations.
-
These validation results merely show how well the created profile models the actual displayed results. They say nothing about the accuracy of the calibration ... though, in general, the more linear your monitor the lower the deviations.
[a href=\"index.php?act=findpost&pid=79752\"][{POST_SNAPBACK}][/a]
so,if delta E test not show accuracy of the calibration,then which way to show?
and why 2 software generate a profile to my display card LUT after hardware calibration?
-
so,if delta E test not show accuracy of the calibration,then which way to show?
and why 2 software generate a profile to my display card LUT after hardware calibration?
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=79754\")
To test gray balance, simply display a gray ramp and look for impurities. You can use the White_Balance.tif available for download from here:
[a href=\"http://www.hutchcolor.com/Images_and_targets.html]http://www.hutchcolor.com/Images_and_targets.html[/url]
I'm not sure why it's updating the video card's LUT. It looks like it's going for maximum colour accuracy at the expense of gray balance.
-
thx you ref image site
-
I've just calibrated my Eizo CE210W with BasICColor 4.02, and the resultant profile is decidedly magenta tinted, both to my eyes and when compared with a grey ramp printed from a well-calibrated printer.
The profile produced by the included Eizo calbration software was even worse, again heavy on the magenta.
Also, when I ran an immediate profile validation from within BasICColor I got the attached - pretty lousy result, it seems to me. I'd be grateful for any insights / advice!
Peter
-
What is the track record on Basic-Color? If it isn't good, perhaps buy a good software package such as ColorEyes Display with an X-Rite colorimeter (www.integrated-color.com). Or if the calibration package isn't the problem, sounds as if your monitor or video card may have issues.
-
Mark, I posted the same issue in another topic (http://luminous-landscape.com/forum/index.php?showtopic=12949&st=20), I haven't got anywhere with this yet.
I do use the X-Rite puck (DTP-94) which is apparently well-regarded. I had some issues with the operation of ColorEyes on my system which I couldn't resolve even with the help of Jack Bingham. So I bought BasICColor which, for my two ACD monitors, gave me a pretty good profile. I bought the Eizo because one of the ACD's died, and after re-calibrating the remaining ACD it still comes up OK.
The bit which is causing me to think something's going wrong in the calibration process itself (as opposed to the hardware), is the poor validation result. I realise that the logic is flawed here, but it seems to me that if the system is actually measuring and displaying the issues I see visually, then the puck at least must be performing OK.
I have to leave it now for a few days, unfortunately - the day job has a habit of intruding on my real life and I have to travel again!
Cheers,
Peter
-
Peter, sounds to me if the puck seems to be operating consistently with expectations and the ACD monitor calibrated and profiled OK, then by process of elimination it would seem the problem is with the Eizo monitor.
-
To be frank, I'd be more inclined to think the problem lies with the idiot trying to profile it! I can't help feeling that there's something I'm not doing right. When I get back from my trip I'm going to try several different softwares and see if I get consistent issues. When I've eliminated my own incompetence and if the problem is still there, I'll return it to Eizo.
-
so,if delta E test not show accuracy of the calibration,then which way to show?
and why 2 software generate a profile to my display card LUT after hardware calibration?
[a href=\"index.php?act=findpost&pid=79754\"][{POST_SNAPBACK}][/a]
As stated, it's simply to inform you of the delta between the various measured processes, NOT the accuracy of the calibration. It can't (no product can) unless you're using a more precise reference device. We've discussed this elsewhere on this site. In a nutshell, if you have a ruler and you use it to gauge the accuracy of a measurement, you're not doing much other than seeing how each measurement might vary from the next. You can't gauge the accuracy unless you use a far more precise ruler and compare that to the original ruler. IOW, using the same instrument to measure the accuracy of that device is folly.
-
As stated, it's simply to inform you of the delta between the various measured processes, NOT the accuracy of the calibration. It can't (no product can) unless you're using a more precise reference device. We've discussed this elsewhere on this site. In a nutshell, if you have a ruler and you use it to gauge the accuracy of a measurement, you're not doing much other than seeing how each measurement might vary from the next. You can't gauge the accuracy unless you use a far more precise ruler and compare that to the original ruler. IOW, using the same instrument to measure the accuracy of that device is folly.
[a href=\"index.php?act=findpost&pid=84929\"][{POST_SNAPBACK}][/a]
Andrew - makes sense - but let me ask you if this is also a correct way of looking at the matter: the software (and the example in my mind is ColorEyes Display because that is the - excellent - one I am using) has a reference file of numbers describing each patch the colorimeter is reading. After doing the calibration and profiling, there is a further "verification" step which reads another 15 or so patches and produces a Delta-e graph. My understanding is that these Delta-e's are a measurement of the difference between the results of the calibration/profiling (as embedded in the new profile) and the numbers in the software reference file describing what those patches should be. Is that correct? Of course this measurement can only be as accurate as the colorimeter allows them to be, which is what I understand to be the point you make about the accuracy of the ruler.
-
Andrew - makes sense - but let me ask you if this is also a correct way of looking at the matter: the software (and the example in my mind is ColorEyes Display because that is the - excellent - one I am using) has a reference file of numbers describing each patch the colorimeter is reading. After doing the calibration and profiling, there is a further "verification" step which reads another 15 or so patches and produces a Delta-e graph. My understanding is that these Delta-e's are a measurement of the difference between the results of the calibration/profiling (as embedded in the new profile) and the numbers in the software reference file describing what those patches should be. Is that correct? Of course this measurement can only be as accurate as the colorimeter allows them to be, which is what I understand to be the point you make about the accuracy of the ruler.
[a href=\"index.php?act=findpost&pid=84930\"][{POST_SNAPBACK}][/a]
I don't know what the Color Eyes product is trying to do but lots of products provide a validation whereby they show you the reference and measured colors and attempt to provide a delta. Frankly, unless the idea is to inform you that it's time to recalibrate, the data isn't very useful based on the points raised above. IOW, say you calibrate your display and build a profile on Monday and on Friday, a client is coming to work with you. You want to ensure the display is in calibration and you don't want to take 10 minutes to undergo the full calibration and profile process. A good validation process will simply compare the older and newer values and let you know if you're good to go or should recalibrate. But if a software product tells you that the process in some way gauges the accuracy of the calibration without sending you a nice $20K Spectroradiometer, they are selling you a bridge as Jack would say.
It's possible that AFTER calibration and profiling, a small subset of patches could be read to allow the software to 'know' no further adjustments are necessary. But that again has nothing to do with telling you that your calibration is where it really needs to be since the entire process is only as accurate as the instrument (assuming no software bugs). So a product may by default ask to measure 100 colors to build the profile and after validation decide it wants to measure another 100 patches because this validation has proven not enough measurements where initially taken. But again, this isn't at all useful for providing a delta of how accurate the entire process is based on a more precise set of measurements.
-
I'm not sure why it's updating the video card's LUT. It looks like it's going for maximum colour accuracy at the expense of gray balance.
[a href=\"index.php?act=findpost&pid=79759\"][{POST_SNAPBACK}][/a]
Unfortunately I don't have a hardware calibratable monitor (yet ) but the only reason I can think of is that the Eizo calibration software updates the Videocards LUT to be sure it's linear, that means that it doesn't tweak the RGB values from the operating system/photoshop.
a LUT is nothing more then a translation table. It translates R,G,B, to R', G',B' were R',G', B' are corrected values for proper colour display.
Below an example from my current monitor profile (icc profile opened using MS Wordpad):
BEGIN_DATA
# ID R G B X Y Z
0 1 0 0 17.329 9.34629 0.927452
1 0.875 0 0 12.6458 6.8302 0.69593
2 0.75 0 0 8.80392 4.76363 0.505212
3 0.625 0 0 5.78843 3.14364 0.357655
4 0.5 0 0 3.51079 1.91771 0.246879
5 0.375 0 0 1.89861 1.0508 0.168689
6 0.25 0 0 0.84074 0.481617 0.118747
7 0.125 0 0 0.265462 0.172215 0.0912377
8 0 1 0 11.1763 23.8853 4.5426
9 0 0.875 0 8.33561 17.7644 3.36489
10 0 0.75 0 5.932 12.5979 2.37949
11 0 0.625 0 3.97122 8.39765 1.59463
12 0 0.5 0 2.44345 5.13725 0.992812
13 0 0.375 0 1.33741 2.77586 0.564622
14 0 0.25 0 0.601686 1.21134 0.283442
15 0 0.125 0 0.197463 0.353186 0.130257
16 0 0 1 6.89753 3.14656 36.0932
17 0 0 0.875 5.12829 2.35846 26.7932
18 0 0 0.75 3.63884 1.68712 18.9491
19 0 0 0.625 2.44373 1.14712 12.6574
20 0 0 0.5 1.51485 0.723885 7.7735
21 0 0 0.375 0.839167 0.417401 4.19966
22 0 0 0.25 0.387226 0.21016 1.81825
23 0 0 0.125 0.144998 0.0992423 0.53658
24 1 1 1 34.5609 35.6367 40.7166
25 0.875 0.875 0.875 25.7282 26.6057 30.4467
26 0.75 0.75 0.75 18.1335 18.8122 21.5683
27 0.625 0.625 0.625 12.0727 12.5571 14.4279
28 0.5 0.5 0.5 7.34231 7.64457 8.81315
29 0.375 0.375 0.375 3.96559 4.13023 4.76694
30 0.25 0.25 0.25 1.72005 1.78928 2.07344
31 0.125 0.125 0.125 0.491885 0.504968 0.598622
32 0 0 0 0.0602451 0.0612967 0.0806908
END_DATA
If a profile like above is loaded into the Videocards LUT, then it does affect the monitor. My profile is loaded by the Monaco Optix software that is executed from the "Windows->Startup" folder when my system starts up. When using a monitor that is hardware calibrated, there is no need to load anything into the Video cards LUT, except maybe a dummy profile to be sure the LUT is neutral. I am not sure but I expect that the Videcards LUT is reset, everytime you restart the system.
Maybe you should check if there is any calibrations software active other then Eizo's own software.
-
There are several points to consider here without making validation valueless. First I find validation is a great tool fom tracking the changes over multiple profiles. It may not be a the best way to gauge the accuracy of a single profile, but it is very effective for evaluating trends from profile to profile. I frequently use it to test various sets of settings against one another. I would not suggest that using it at a later date to verify the condition of the monitor/profile is valid because of non-uniformity across most screens. The likelyhood of placing the instrument in exactly the same place is a real long shot. Validations made at the time of profiling and stored for reevaluation are far more valuable. Finally while being limited to evaluating a profile made from your instrument, through your instrument which may diminish the absolute value of the validation I still feel there is valuable information there. Godpel it's not. A reasonbale snapshot it is, unless of course your instrument is way out of whack.