Grayscale variances - different patterns, plasma calibration - Page 7

David Abrams, from Avical, tested the patterns across multiple displays while we were developing the equal energy patterns. They apply to gray scale, gamma, and gamut. They help plasma as well as a display that uses a dynamic IRIS.

Assuming a display gamma=2.2 what will be the constant average luminance value of these patterns as a percentage of peak white?

I'm not aware of Sony ever making plasmas, so I'll guess you're probably not talking about what most people would term ABL. Sony has made displays with things like a dynamic backlight or an adjusting iris, and since you mention full-screen (fields) I'll figure that's the sort of thing you're probably talking about. If you're looking at a different display technology, the expected measurement variations are probably different from plasma. For example I think at some point someone posted some measurement patterns that held a constant average video level and apparently measured similarly on plasma, but the patterns caused the auto iris on my display to adjust significantly.
Edited by alluringreality - 2/24/13 at 7:33pm

I'm was only bringing up the example because of the behavior seeming similar, as the dynamic backlight was off for calibration on the hx850 I was calibrating. Yet, I saw the gamma response behave somewhat akin to what was being discussed related to ABL light response.. I would too like a solution/technique for auto-iris impact and plasma displays ABL for the displays I own/calibrate. This is a great thread... If I don't have to deactivate Dyn BL and still get linear light for calibration, it would serve all... At this point small patterns seemingly are the solution for some, but then here was a comment (chad b) that discussed, I believe, a Samsung that didn't behave like a typical plasma related to ABL. So, I started following the discussion. Thanks for helping me not add confusion.

Would these possibly be available for use with the VForge via SD card?

The video forge can already do this. I believe u can choose constant APL (u can also select the % for the APL) and window size.

I was aware and using various window size %s; I use that function quite a bit. But, I was not aware of APL patterns existing in the VForge (current not 4K). Maybe I have a took I didn't realze -wouldn't be the first time:rolleyes:
Thanks,
tbaudoin

What Spears and Munsil appears to be calling "equal energy patterns" would appear to be different than the industry definition for APL. The way I read it, Spears and Munsil has probably assumed a theoretical display gamma and intends to keep average light output constant at a theoretical display. The AVS HD 709 patterns marked "APL" both hold average video level (APL) constant and they also hold a constant average light output at the display during the measurement run. As I meant to imply in my prior reply, constant APL might have a use on plasma, but constant APL probably has similar problems to typical windows for gamma measurements on displays with something like a dynamic backlight or iris. I would guess that "equal energy patterns" may provide better gamma correlation for measurements across a variety of display types and settings than typical windows, but I wouldn't expect "equal energy patterns" to provide any better gamma measurement correlation than if everyone was using something like the APL pattern from Chad B.

Thanks, it can get convoluted. It's great that we are all trying to get on the same page for discussion purposes. Again, thanks..

As far as the VForge, I absolutely didn't realize it had APL available. I just looked at the website and found the following:
Version 447 (14Oct10)
1.Added "Constant APL" option to the Window pattern.

So, now all I have to do is find the way to activate it. If you happen to know off the top of your head, I would be grateful.

Regards,
tbaudoin

Anyone have the APL % working with the VForge? I was reading that those were added in Oct 10, but according to a note by Chad on SpectraCal's Forum, he couldn't use them except manually (little more involved than this, but I thought I would add this for discussion).

T.

Also added to the Accupel 5000 pattern generator in the Sept. 2012 firmware release..

Also

The video forge does have some built-in APL patterns that are now accessible via the pattern window size in CalMAN 5.1

Just for reference, the GCD disk "10% windowed APL" patterns have a constant average video level (stimulus) of 22%. The total light output assuming a 2.2 gamma ranges from 3.7% to 11.1% of peak derived output for the gray scale/gamma sequence. The most stable gamma response from my previous measurements (samsung plasma device) is achieved when using patterns with total light output in the range of 3% to 21% of a peak output when set to 120 cd/m^2

Yes I've been enjoying them. They are proving very useful with the Panasonic 50 series.

Actually it's very easy with a K10-A, which it the probe we use for plasmas, and the reason for suggesting it is we use it all the time with very, very good results in professional post-production operations where they have not yet replaced the plasmas they purchased before the issues with ABL were fully understood...

I would never suggest something we do not use as an integral part of our calibration process.

But, as I keep saying, if you have a process that works for you, stick with it.
With plasmas there really is little that is truly right or wrong.

Steve

I thought some real-world viewing data might be interesting to discuss so I measured the average luminance of the display over as large an area as possible. I took a look at two examples, the first was an episode of "Justified" which leans heavily on dark scenes. And the second was the first hour of "The Grand Hotel", for typical hollywood balanced, high contrast lighting (all interiors).









All of the images are well below where the ABL has any significant effect on the gamma response on my display. The mean and median for Justified was about half that of Grand Hotel, which was ~8%. This corresponds to stimulus levels in the range 23%-32% for a 2.2 gamma display.

edit: I added indicators of the luminance operating points the AVSHD and GCD APL patterns will calibrate. Also noted is the location a grayscale ramp will generate (I verified this via measurement as well) and I would not recommend using a pattern derived from these levels.
Edited by zoyd - 2/26/13 at 5:57pm

If you're using a plasma it may complicate the data compared to a more ideal display, but that probably doesn't matter for a basic look at things. You're also only measuring a portion of the screen, which again may or may not matter. I've never seen either video, but generally I would have guessed a little higher (between 10%-15%) based on the video levels that I've done a quick gamma adjustment and average on. Of course with the DVE website stating only 15% for APL then something like 8% or lower for average luminance wouldn't be unreasonable.

I don't agree with the exact locations, since for example a lot of the information for the AVS APL patterns is at the edge of the screen, but the general locations are expected. The AVS small APL was intended to be darker than average content and the AVS large APL was intended to be brighter than average content. The APL pattern series from Chad B is probably more in line with what most people are interested in with this sort of thing, and the Chad B APL would likely make a more sensible comparison against something like the GCD APL. The measurements from Chad B suggest that measuring a pattern with a static APL and a constant average luminance (Chad B APL) is probably not the same as measuring a static APL-only pattern (GCD APL) on plasma, so I have some doubts how well a theoretical constant average luminance (equal energy) pattern will hold up in actual practice across a range of display types and settings. Based on the measurements from Chad B I'm open to the possibility that it may not make sense to use something like the Chad B APL for adjusting RGB balance on plasma, but I still can't come up with a reason why for the majority of displays it might prove an advantage to compare gamma measurements using patterns that vary average luminance (windows, fields, static APL-only) or patterns that vary APL (equal energy).
Edited by alluringreality - 2/26/13 at 10:53am

Light Illusion
Could you specify the size of the pattern you received on your technique for plasma Panasonic 50VT50E ?

Actually, for the predominant range of luminance in the measurements, this plasma behaves pretty close to ideal.

For your AVSHD patterns I scanned the probe across the horizontal extent of the screen and averaged a left side, middle, and right side view to capture the edge information so they should be pretty close to theoretical. Grand Hotel is 4x3 so I was capturing most of those images. As I said, this movie is entirely made up of interior shots so I would expect a shift to higher averages, maybe in the 10-15% range, for movies with a predominance of daylight scenes. In any case, the gray scale ramp method of setting a "typical" display luminance for pattern generation is not in-family with typical movie levels which is what got me thinking about doing these measurements in the first place.
Edited by zoyd - 2/26/13 at 10:53am

From a couple quick checks I still think applying a curve to an image is the same as a power function, so there would seem to be variation beyond a simple difference in display gamma. The following were my estimates for 2.2 gamma:
AVS L APL - 25% average luminance
Chad B's APL - 13% average luminance
AVS S APL - 5% average luminance
Edited by alluringreality - 2/26/13 at 4:07pm

There has been no single size.
It varies for each display we have profiled.

That is the idea of the procedure - you define the ideal patch size foe any given display.

Steve

At least approximately - 3 - 7%?

you take the diagonal diameter or height or width TV?

K10-A diagonal (my TV) distance - 685 cm .

D = 2L tan(θ/2)
TLT (Degrees) for K10-A 10.6

If I had to give a figure under pain of death, I would say 5% to 15% of screen height...

window size ?

If the pattern is square, I'm thinking that's less than 2% area on 16:9.

If I'm doing my own patterns, I do them in 16:9?
Area 2% ?

yes, I figured out I can't measure your APL pattern luminances the way I was doing (undersample spatially and then average across the screen) because it weights the center patch too heavily. I just measured the patch size and calculated the following: 24% for the L APL and 4% for the S APL

Here is an hour long sample of Tree of Life which is almost entirely shot with natural light. Avg= 14%, median=11.9%