Benq SW2700PT Palette Master Element SW - absolute vs. relative black point

[rasworth]

Recently purchased the SW2700PT, using the Palette Master Element SW for calibration/profiling.  It appears to be a severely dumbed-down version of i1Profiler, required in order to set the monitor LUT.  All makes sense, except for the black point choice, Relative vs. Absolute.  I had assumed that Relative would allow me to input a contrast ratio, but no numeric field was presented.  Ran passes both ways with all other parameters set the same, the results appear identical.  I'm using an i1Pro for the measuring instrument.

I called Benq tech support, they had no idea as to its meaning.  Anybody in the forum provide help?

Thanks,

Richard Southworth

[rasworth]

After some googling I'm guessing it specifies the rendering intent, and indeed ColorThink shows different chromaticity values for the black points in the two profiles.  I'm still confused, why would one do other than relative?

Richard Southworth

[rasworth]

Still looking for wisdom, anybody?  I do have a correction, the choices for black point are "Relative" or "Absolute zero".

Richard Southworth

[JNB_Rare]

The reviewer in the following article mentions that he used 'absolute' when setting up with Palette Master, but he doesn't mention why. He has a few other conclusions, one of which is that the regular i1Dispay Pro software produced better results (in his opinion) than the hardware calibration with Palette Master.

SW2700PT Review

[Royce Howland]

In this context, relative or absolute has nothing to do with rendering intent. RI is not applicable to monitor calibration. Rather, it has to do with how the software is setting the monitor's black point. I downloaded the BenQ Palette Master software manual, but it's fairly useless. Just based on knowing how these applications are designed to work, with "absolute zero" the software should be targeting a specific brightness level for the black point, presumably 0 cd/m2. (Or whatever the actual darkest black point the monitor is capable of hitting.) With "relative" the software is likely targeting a black level that falls at some point relatively determined by the white point brightness and contrast ratio. But it seems like you're not able to specify a contrast ratio in the interface, so who knows what this is really doing.

I'd say use the "absolute zero" setting. I normally want a monitor to deliver the deepest black it's reliably capable of, rather than trying to hit a black level based on some arbitrary choice of contrast ratio.

[rasworth]

Thanks for the link, I thought I had dug out all of the reviews but I missed this one.  The problem I see with using i1 Profiler is its load of the 8 bit video adapter LUT, as opposed to the 14 bit LUT in the monitor, only accessible with the Palette Master Element software.  I may have to do some experimentation, using i1Profiler after Palette Master, perhaps the combination will let me create a better profile and still retain a "straight line" load of the video adapter LUT.  And of course still no definition of "Relative" vs. "Absolute Zero".

The profile generated by the included SW has a vcgt tag, but there is no provision to load into the video card, i.e. no startup program as with i1Profiler (which I had to disable).  And so far I have found nobody in Benq that knows anything about the SW, I suspect it was contracted out to Xrite or similar.  Outside of the calibration/profiling mystery it is a very satisfactory wide gamut display, and I have not detected any homogeneity issues.

Richard Southworth

[rasworth]

Royce,

Rendering intent was a bad choice of terms, I thought perhaps the Absolute Zero choice would "bend" the neutral axis from the chosen white point at the bright end to some other choice at near zero.  However I was not able to visually detect any difference using test images.  And I also assumed Relative would work to set a contrast ratio, except as you pointed out there is no provision for such.

Richard Southworth

[rasworth]

I went back and examined the media black point xyY values for the Relative vs. Absolute Zero profiles, the Y values were close to equal, although there was significant variation in the xy chromaticity coordinates.  I am using an i1Pro spectro for these measurements, which is not particularly accurate at the dark end.  Guess I'll have to spring for an i1 Display Pro for more accuracy.

Richard Southworth

[Erland]

I thought absolute zero was the device 0,0,0 in RGB values and relative a more neutral black target based on measurements?

[rasworth]

I thought absolute zero was the device 0,0,0 in RGB values and relative a more neutral black target based on measurements?

Makes sense to me, although it may be difficult to verify, I assume the adjustment would be made in the monitor LUT, which as far as I know can't be examined.  I have an i1 Display Pro coming, maybe I can tell from the profile data.

Richard Southworth

[rasworth]

Received the final answer from Benq wrt "Relative" and "Absolute Zero":

"We currently don’t have documents on this feature. But I can provide a brief description on the two options:
•   Absolute: using ‘0’ as the black point measurement, so it gives you the best contrast out of the monitor. (It basically sets the black of the panel as black point.)
•   Relative: using the gamma curve to calculate the suitable brightness of the black point. The black is not necessary the darkest brightness level of the panel, but will give a better fit to the target gamma curve. This is often used in soft proofing application.
Also, the black point options will be recorded in the ICC Profile generated after calibration. It will also affect the method of ICC profile conversion in Photoshop or other software application."

After many passes with my new i1 Display Pro, here's where I ended up:

1. Used the Benq sw, "Palette Master Element", to calibrate/profile at my desired WP, Luminance, and Gamma.  This loads the 14bit LUT in the monitor as well as creating a profile.  No load into the video card LUT is necessary, although the profile does include a vcgt tag, I had to disable the calibration loader to prevent unwanted data from loading.  The resulting profile was usable although a little "raw", was not of a format that would allow QA runs from i1Profiler, and I believe had some errors in non-essential tags.  There is no provision for contrast ratio.

2. Used i1Profiler to generate a new profile, using native settings for contrast, WP, and luminance.  Unfortunately there is no provision for native on gamma, so used the same value as with the Benq sw, i.e. 2.2.  The resulting vcgt tag loaded what appeared to be a straight line into the video card.  I ran QA with and without the vcgt tag loaded, and the results were virtually identical, so I disabled the calibration loader.

I now have what I believe is the best of both worlds, a properly calibrated monitor at my desired WP/lum/gamma, and a decent profile.  The final contrast ratio was approx. 1100, no way to adjust since it would require a different LUT load into the monitor, so the only way to get realistic black levels is to softproof within Photoshop.

Richard Southworth

[Royce Howland]

From their description, it seems like the Palette Master Elements software is doing what it would be expected to do. Given that, I think your workflow is reasonable to get the best of both worlds. You're essentially using BenQ's proprietary app to calibrate, or linearize, the monitor hardware via its on-board LUT. Then using i1Profiler to generate the colour profile on top of that, which will be used for colour transformations in Photoshop and other colour managed applications.

If you're ever curious whether there's a difference in display accuracy stopping just with the Palette Master-generated profile vs. going the extra step to re-profile with i1Profiler, other than just going subjectively by which looks better to your eye, there are ways you can objectively check it out. BabelColor makes a nifty application called PatchTool, with a function called DisplayCheck. It runs a verification of various user-selectable patch sets through a display profile, and provides some comprehensive reporting on the results. You could run DisplayCheck with the Palette Master profile, then again with the i1Profiler profile, and see whether there's any significant difference between the accuracy reports.

http://www.babelcolor.com/patchtool_display-check.htm

Of course there are also profile analysis tools like ColorThink and GamutVision, as well as tools within ArgyllCMS, any of which might illustrate differences going on within the profile itself. But while I use those tools, I also like the BabelColor DisplayCheck tool because it actually runs live end-to-end through the display pipeline. This ensures that everything that one thinks should be working a certain way in fact is working.

[rasworth]

I have ColorThink Pro, used it to examine the Benq profiles.  DisplayCheck looks interesting, but I'm not anxious to spend the bucks at this point, particularly since the i1Profiler profile seems to work well.  I did a Quality run in i1Profiler, using the 928 patch included test set, GRACol 2006 Coated 1.  I've attached the results summary, perhaps somebody can tell me if this represents goodness.  All visual checks (neutral step charts, test images, etc.) are positive, at least to my eyes.

Richard Southworth