Calibration tests with an EyeOne D2 [Warning: extremely nerdy]

Ever since this discussion back in February, I've been meaning to run some calibration tests. This will be totally boring to most people, but I hope at least Lowner and gcogger may read it

(I'll have to break this up into several posts, because of the image limit.)

I used my trusty Eye One Display 2, and my guinea pig was a run-of-the-mill LG Flatron L1919S TN panel monitor (coupled with run-of-the-mill onboard graphics - SiS 661FX if that means anything to anybody).

The Prep

I wanted to try various calibration targets, mainly to assess banding. Monitor banding is so feared among photographers, but I've never seen it, so I wanted to find some. I felt like Captain Ahab seeking Moby Dick.

My monitor's factory defaults are:
Brightness 100%
Contrast 70%
Red, Green and Blue (when activated) 50%

Now, I had a list of various targets and methods I wanted to try (see below), but of course I couldn't try every single possible combination. So I decided to
(a) stick to Luminance target of 100, which has always served me well.
(b) follow the EyeOne's lead on the Contrast setting each time. I have no quibble with Contrast (at least, I thought not - more about that later)

Test 1 - Targets: Native/Native/No change

Targets: Native white, Native gamma, No change luminance (let's start by seeing just how bright this baby can be)

Method: Started with factory defaults (100, 70, 50/50/50) then took Contrast up to 100 per EyeOne's instructions.

Resultant monitor settings: Contrast 73, all others unchanged.

Calibration result: 6100K, 257cd/m2 - damn that's bright!

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

Test 5 - 6500/2.2/100

Method: Set the target to 6500, and actually used the monitors presets to get as close as possible this time. As it turned out, the 6500 preset was too low, but the next highest preset (unnamed) was 6500 on the nose. Just lucky, I guess.

Resultant monitor settings: Contrast 73, Brightness 4. Two points higher in the Brightness this time, not quite sure why.

Calibration result: 6500/2.2/86. Hit the white target, but again the Luminance was low. And look at that splayed Gray Ramp! Yeesh!

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

Test 9 - 6500/2.2/100

Method: This time, I started with the R,G & B at 50s, with the aim of keeping the Brightness, R, G and B settings all within sight of each other. This has always been my preferred approach, although I've never had any scientific reason to do so, only my gut instinct based on many many calibrations.

Resultant monitor settings: Contrast 73, Brightness 23, Red 22, Green 28, Blue 17. It took me a couple of attempts to get this right - I had to jump back and forth between the White setting and the Luminance setting.

Calibration result: 6500/2.2/99 - best one yet.

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

HTTP response: 404 | MIME changed to 'text/html' | Byte size: ZERO

I don't completly grasp this subject and calibrated mine according to generally accepted practices. But After reading this I may have to go have a play of my own sometime.

When looking for banding I prefer a "Granger rainbow"...I have some banding on the rainbow but have yet to have it be an issue in the real world.

I found your variations on the grey bar of special intrest...I have never really cared for how my grey's appeared after calibration but had decided to trust xrite over my perception. I cannot even say there off one way or the other....There just "Blah", for lack of a better discription.

Interesting read.
I'll read it again in a while, and that time engage the brain as well. Now it's weekend

neumanns wrote in post #7806198
When looking for banding I prefer a "Granger rainbow"...I have some banding on the rainbow but have yet to have it be an issue in the real world.

I found your variations on the grey bar of special intrest...I have never really cared for how my grey's appeared after calibration but had decided to trust xrite over my perception.

Interesting that you appear to pay more attention to colour than greys. I've always found greys to be the true test of profiling, either display or print.

A granger rainbow would have been of little use to me, frankly. I forgot to mention it above, but I could never see any banding in the colour gradients in my test file. It was only visible in the grey gradient. I was only using the colours to check hue.

The thing with the color granger that I found interesting is there is a band that is visable that is not visable in the greyscale.

My banding appears wherever one or more RGB values are 254 or 255 range, This lays a predictable grid across my granger rainbow in both directions. My assumption is this has to do with the hilight ramp.

However the horizontal band all but dissapears in greyscale at the same 254/255 value's

Damien,

Way over my skill level, but I promise going to print it out and then do some serious studying. Thank you for what must have been a lot of work. Is there a test to see what I've actually retained?

You commented in test 1 about the blue looking greys at Native. That odd to me, because left to its own devices, my Samsung Syncmaster 910n has a distinct pink edge. I've solved that by using 7000 deg K which is higher than the recommended 6000 - 6500.

Its also a relief to see how similar your input/output graph is to mine. I've always fretted that the blue seems so far out of step with the red and green, particularly at the 255 end of the scale.

Damien,
Of course, I don't use the EyeOne, but instead utilize the much maligned Spyder3. But your testing has encouraged me to try your method and recalibrate my Dell 2209WA using the RGB sliders.

Lowner wrote in post #7811082
Its also a relief to see how similar your input/output graph is to mine. I've always fretted that the blue seems so far out of step with the red and green, particularly at the 255 end of the scale.

Richard, perhaps you've misread. I, too, have always been very precious about my Gray Ramp. I was surprised, I guess, to find it made very little visible difference to the calibration.

That said, I'm still going to aim for a straight, close-together ramp as seen in tests 8 and 9.

I wish I'd taken a moment to compare the difference between Photoshop and a non-colour-managed viewer in each test. My guess would be that a "good Gray Ramp" profile (Tests 8 & 9) would give closer CM-to-non-CM results than a "splayed Gray Ramp" profile (Tests 5 & 7), because more of the adjustment existed in the monitor settings than the profile. I might go back and test that, if it's possible to do so.

ChasP505 wrote in post #7813245
Damien,
Of course, I don't use the EyeOne, but instead utilize the much maligned Spyder3. But your testing has encouraged me to try your method and recalibrate my Dell 2209WA using the RGB sliders.

I understood that the Spyder3 was much less maligned than its predecessors. In any case, I'd be really interested to hear your results.

Damien,

Enboldened by your testing, I've just recalibrated using the RBG channel sliders. It took me three attempts, and I've had to bring the colour temperature back down to 6500 because my previous 7000 just looked mad, but the graph is much improved. Still a little wayward, but closer than I've ever seen them before.

My settings:
Brightness: 95
Contrast: 82
Red: 31, Green: 9, Blue: 43

That's calibrating to 6500/2.2/110

Wow, that Green is low, isn't it? What were your starting points for R,G & B? 50s?

Damien,
I sat through a color management video tutorial last night by Chris Murphy, co-author of Real World Color Management. He's got some interesting philosophies on calibration, often totally counter to the common recommendations given on this forum.

He advocates a work/lighting environment which meets D50/5000K conditions. Following from that, he says that calibrating to D50/5000K is best, but acknowledges that many can't keep their work environment at those conditions, so 6500K is next best.

He ran through an LCD display calibration with the EyeOne 2, targeting 120 for the white luminance, 6500K, and 2.2 gamma. Interestingly, he adamantly says to NOT use the RGB controls and to ONLY use the actual identified backlight control, whether it's called Backlight, Brightness, Contrast, or whatever is the case for your respective LCD display. He also said to not use ambient light measurement or compensation, opting for tighter and more consistent control of your image editing work environment.

I followed his prescription with my Spyder3 Elite and Dell 2209WA and, at least subjectively, the result can be described as the "smoothest" I've gotten so far, judging from viewing a group of grayscale calibration images in PS CS4.

My display's Contrast setting remains at the factory default setting of 75, and the Brightness (backlight) is at a low setting of 12 to achieve a target of 120cd/m2. The black value is at 0.24 cd/m2. Bottom line for me is that I'm now totally convinced (and persuaded) that Native is NOT the way to go, but I'll still not use the RGB controls when calibrating the display.

Damien,

Yes, the green channel amazes me too! I did not reset the screen to either the defaults or 100, or even zero. That's for when I've digested your report and try again. For now I just started where the menu had set the channels. They started out at:

Red:50, Green:40, Blue:60

However that was with the previous 7000 degK colour temp, so reducing it back down to 6500 degK will have disturbed things. I'm amazed that A. I've had to use 7000 up to now, and B. I've now been forced back to sensible levels just because I used the RGB channel adjustments for the first time ever.

But this is not my last go at it. Your hard work has spurred me into doing a similar amount of work on mine with hopefully a similar outcome.

Glad I inspired you, mate! I'd be particularly interested to hear the results if you tried Test 8 (starting with RG&B all at 0) and Test 9 (starting all at 50).