[zoyd]

I got this idea from the patterns described in the display metrology document from this thread. They can be used to measure your display minimum luminance level (MLL) and see how it varies with the average picture level (APL). Full field black and ANSI patterns may not capture the idiosyncrasies of your display that can affect contrast ratio when viewing real content. There are 13 levels on the disk which contain a central 10% area black window surrounded by a random star field pattern. The star field pattern contains at least 10% of the pixels distributed between levels 16-235 so that all video levels will be stimulated at each pattern level even lthough the average will vary as given below:

1 0%
2 4.8%
3 9.3%
4 13.7%
5 20.4%
6 24.8%
7 29.3%
8 33.7%
9 38.1%
10 51.3%
11 61.6%
12 72.8%
13 87.5%


Using these patterns to measure MLL vs. APL on my display, all measurements using a D3:



You will notice that at 10% average stimulus the D8000 MLL rises more than a factor of 2 and then tapers down slowly. The ANSI measurement is also shown and it has not budged from the full black level. So if all you did was measure full black contrast and ANSI contrast on this display you would overestimate it's contrast performance when displaying real video content. It would be interesting to see other panels measured in this way.


updated 06/14
-added asymmetric patches for local dimming tests
-added 3x3, 4x4, 5x5, 6x6, 12x12 ANSI patterns
-added Dynamic contrast test (25% fixed on variable background)


updated 06/18
-added ABL loading patterns, peak white on variable background up to 97% average stimulus
-reduced disk size, patterns run for 10 seconds each

The patterns in AVCHD/DVD format can be downloaded here.

 

[TomHuffman]

I am afraid that this is a Samsung-only phenomenon that you have measured. I tried this with the Pioneer and the MLL stayed below the measurable threshold throughout the entire range. Then I tried with a Panasonic ST30. The luminance stayed almost exactly the same at all levels, including the checkerboard.


APL.................x....................y....................Y

Black...............0.281.............0.272.............0.0236

Level1.............0.290.............0.271.............0.0242

Level2.............0.289.............0.272.............0.0243

Level3.............0.289.............0.273.............0.0243

Level4.............0.290.............0.271.............0.0241

Level5.............0.290.............0.272.............0.0241

Level6.............0.291.............0.272.............0.0241

Level7.............0.290.............0.273.............0.0241

Level8.............0.289.............0.272.............0.0239

Level9.............0.290.............0.274.............0.0239

Level10...........0.291.............0.274.............0.0238

Level11...........0.289.............0.273.............0.0236

Level12...........0.292.............0.273.............0.0236

Chckrbrd.........0.287.............0.273.............0.0244

 

[zoyd]

Tom, thanks for trying it out. I expect the Panasonics from this and last year are ok, as well as the E-series. Anyone with a good meter want to try it on a Sharp Elite?

 

[D-Nice]

Next time I deal with one of them, I'll provide you with the results.

Quote:
Originally Posted by zoyd  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl/0_100#post_22130239


Tom, thanks for trying it out. I expect the Panasonics from this and last year are ok, as well as the E-series. Anyone with a good meter want to try it on a Sharp Elite?

 

[AnotherDude]

Elite 70X5 Local Dimming On




Meter: X-Rite i1 Display Pro Retail (not super sensitive ... see those 0s)

ISF Mode

Calibrated Gamma = 2.3


Backlight = 30

Contrast = 79

Brightness = 1

Gamma = -4


APL....................x....................y....................Y

Black...............0.000.............0.000.............0.0000

Level1.............0.000.............0.000.............0.0000

Level2.............0.000.............0.000.............0.0000

Level3.............0.000.............0.000.............0.0000

Level4.............0.000.............0.000.............0.0000

Level5.............0.000.............0.000.............0.0000

Level6.............0.000.............0.000.............0.0000

Level7.............0.000.............0.000.............0.0000

Level8.............0.000.............0.000.............0.0000

Level9.............0.174.............0.0075............0.0001

Level10...........0.174.............0.0075............0.0001

Level11...........0.341.............0.088.............0.0022

Level12...........0.292.............0.273.............0.0045


Using the 16-block 4x4 (8 white/8 black) in the AVSHD Disc ChromaPure ANSI Contrast pattern:


1-2-3-4

5-6-7-8

9-10-11-12

13-14-15-16


Chckrbrd 9........0.231.............0.188.............0.0075

Chckrbrd 11......0.234.............0.189.............0.0072

Chckrbrd 16......0.239.............0.192.............0.0063

 

[zoyd]



Quote:
Originally Posted by D-Nice  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl#post_22130685


Next time I deal with one of them, I'll provide you with the results.

Thanks! I am thinking of adding another black patch area that will be more of a torture test for local dimming sets, the 10% central box is probably fairly easy for them to handle. Do you have any recommendations on shape or orientation? I've heard the Elite has 200 local dimming zones (don't know if that's true) but if distributed in equal area boxes that's roughly 100x100 pixels for each zone. I was thinking of a circular patch with radius of 100 pixels to ensure that it would cut across zones and be big enough for a contact meter. I think I will also add the ANSI patterns, any other suggestions?



@AnotherDude


Thanks, I'm assuming your measurements are ftL, correct? You are seeing the effects of patch size/orientation with respect to dimming zone.

 

[D-Nice]

It could be circular, octagon, diamond or triangular... basically anything that does not adhere to the zone structure.

Quote:
Originally Posted by zoyd  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl/0_100#post_22131382


Thanks! I am thinking of adding another black patch area that will be more of a torture test for local dimming sets, the 10% central box is probably fairly easy for them to handle. Do you have any recommendations on shape or orientation? I've heard the Elite has 200 local dimming zones (don't know if that's true) but if distributed in equal area boxes that's roughly 100x100 pixels for each zone. I was thinking of a circular patch with radius of 100 pixels to ensure that it would cut across zones and be big enough for a contact meter. I think I will also add the ANSI patterns, any other suggestions?

@AnotherDude

Thanks, I'm assuming your measurements are ftL, correct? You are seeing the effects of patch size/orientation with respect to dimming zone.

 

[AnotherDude]

Sorry, I almost forgot about this thread! Just subscribed to it, in case.


Yes, ftL measurements. I'm guessing torturing LCD/LEDs is a hobby for plasma owners

. I look forward to the circular patches. For me it's the end result I really care, though measurements are good for seeing relative changes for my purposes.


When you said I'm seeing the effects of patch size/orientation with respect to dimming zone, what exactly were you refering to? Right now my biggest problem with this exercise is my meter is not all that sensitive

 

[zoyd]



Quote:
Originally Posted by AnotherDude  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl#post_22211397


When you said I'm seeing the effects of patch size/orientation with respect to dimming zone, what exactly were you refering to? Right now my biggest problem with this exercise is my meter is not all that sensitive

The difference between your checkerboard measurements and level 12 of my disk are probably due to the size/orientation of the measured black area. The current version adds a triangle with rectangular base, those should measure higher than rectangular center areas. I haven't gotten around to circles yet.

 

[AnotherDude]

Gotcha.

 

[Make73]

Samsung PS64E8000 FW 2006

i1 Display Pro


% / Y cd/m2

0% = 0.017

4.8% = 0.017

9.3% = 0.019

13.7% = 0.038

20.4% = 0.044

24.8% = 0.051

29.3% = 0.055

33.7% = 0.059

38.1% = 0.063

51.3% = 0.086

61.6% = 0.088

72.8% = 0.087

87.5% = 0.086

% / Y

2.5% = 4.740

7.3% = 4.734

11.8% = 4.624

16.2% = 4.518

22.9% = 4.478

27.3% = 4.460

31.7% = 4.466

36.2% = 4.294

40.6% = 3.809

53.8% = 2.708

64.1% = 2.322

75.3% = 2.008

90% = 1.634


ANSI 3x3 B=0.019 W=65.206 (measured from center of screen)

ANSI 5x5 B=0.023 W=71.208 (measured from center of screen)

ANSI 4x4:

0.019 71.148 0.020 70.9

66.1 0.027 72.7 0.021

0.021 72.9 0.023 68.3

69.9 0.020 72.5 0.020



Anyone knows why Samsung give so high MLL readings with higher APL patterns?

And why I get such a normal readings with 4x4 ANSI pattern where APL is 50%, or is it?

Samsung PS64E8005 200h Movie calibrated.zip 45k .zip file

 

[zoyd]

Thanks for the data, here is the mll plotted.




Compare that to the D series in the first post and it shows the same transition at 10% APL but does not level off like the D series. It would be very interesting to see the same measurements on the F-series with the black optimizer in it's three settings, off, dark room and bright room.


Regarding why you don't see the rise with an ANSI pattern, I can't answer that other than to say that has been the observed behavior in these plasmas and why you shouldn't trust the ANSI reading for a static contrast number.

 

[Make73]

Thanks for the diagram.


btw, sure there´s what you measure with meter but it´s different what you see by eyes

A & B are same levels

 

[zoyd]

Here is a follow up to the post I wrote in the F8500 settings thread showing the static contrast performance of 3 panels, one each in the D, E, and F series using the mll vs. apl patterns, along with the Panasonic VT50 and ST30. Of the Samsungs, the F series is of course significantly better at low APL scenes (a factor of 3 vs. the D8000 and a factor of 2 vs. the E8000 at 2% average luminance). It's surprising however that my tweaked panel outperforms the F8500 (3000 vs. 2500) above 7% average luminance @42ftL. ABL brings it back down to match around 42% average luminance. I should note too that the F series is capable of much higher than 42 ftL peak white where this comparison is made. The panasonics do not modify black level with APL and will therefore maintain a constant contrast ratio until peak white is affected by the brightness limiter, typically above about 30% average luminance.




For some context on average luminance levels here is an example of low vs. typical APL.


Typical - Average luminance is 15.1% on 2.2 gamma display




Low - Average luminance is 3.5% on 2.2 gamma display




Low - Average luminance is 4.2% on 2.2 gamma display




Summary of mll for low and typical luminance ranges:




and contrast ratio:




There are limitations to the perceivable benefits of high native contrast ratios, one of which is environmental lighting. Even in a dimly lit room say with several lux falling on the screen and a good AR coating, the reflected light will be at or above 0.01 ftL So to benefit from a display with mll significantly lower than this the first thing you have to do is turn off all the room lights.


The second limitation is the dynamic range of the eye's sensitivity. At any particular adaptation luminance the highest perceivable contrast ratio is ~1000:1 (some would put this lower, maybe 500:1) So if we take our typical average luminance level for a gamma = 2.2 display calibrated to 35 ftL and divide by the maximum perceivable contrast ratio at that adapted average luminance we get (15%) * (35 ftL) / 1000 = 0.005 ftL (ANSI CR = 7000) . So again, once you get into the mll range of 0.005 - 0.01 ftL you need specific conditions to occur in which you will benefit from an mll lower than this. One is minimal room light and the other is for scenes that are 2 or 3 times below typical average luminance levels.


note: Thanks to buzz, Make73, and Tom H for providing the data

 

[Make73]

I did re-calibration/check at 560 hours of usage with FW 2008.2 and also remeasured with Your disc (three measures with both CS ON and OFF)
(If set cell light and contrast to max 20/100 I can reach 152cd/m2 max luminance without crushing white and seems that affect just a liitle bit to calibration, dE´s still under 3)
Samsung PS64E8005 560h 120cdm2 Movie calibrated Masciors Disc Pana matrix.zip 5k .zip file


Cinema Smooth ON

% / Y cd/m2

0% = 0.018 - 0.019

4.8% = 0.018 - 0.019

9.3% = 0.018 - 0.019

13.7% = 0.034 - 0.036

20.4% = 0.046 - 0.047

24.8% = 0.053 - 0.055

29.3% = 0.059 - 0.060

33.7% = 0.062 - 0.064

38.1% = 0.067 - 0.068

51.3% = 0.084 - 0.087

61.6% = 0.085 - 0.087

72.8% = 0.084 - 0.087

87.5% = 0.084 - 0.086


Cinema Smooth OFF

% / Y cd/m2

0% = 0.017 - 0.018

4.8% = 0.017 - 0.018

9.3% = 0.017 - 0.018

13.7% = 0.032 - 0.034

20.4% = 0.050 - 0.051

24.8% = 0.057 - 0.059

29.3% = 0.062 - 0.064

33.7% = 0.066 - 0.067

38.1% = 0.070 - 0.071

51.3% = 0.090 - 0.091

61.6% = 0.092 - 0.093

72.8% = 0.092 - 0.093

87.5% = 0.092 - 0.093


zoyd, any reports that your trim tweak also works with E-series?

Is there anykind of downside with trimming voltages and timings?

However I´m not sure if I dare to do that

 

[zoyd]



Quote:
Originally Posted by Make73  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl#post_23384081


zoyd, any reports that your trim tweak also works with E-series?

Is there anykind of downside with trimming voltages and timings?

However I´m not sure if I dare to do that

No reports but the controls are available on the 51" models.

your repeat measurements are nice and stable, and CS on does not raise the mll.

 

[todd95008]

I was not aware of this thread but it has some great info...


I was testing MLL last night and found this behavior with a Samsung 51" F8500.
http://www.avsforum.com/t/1475066/official-samsung-pnxxf8500-owners-thread-no-street-price-talk/1530#post_23704586


Zoyd, you mention some tweaks you did to a D8000 Samsung, are those tweak documented somewhere on AVS Forum ??

How much of change did they make ?


Thanks


Todd

 

[PlasmaPZ80U]

that floating black level on the Samsung F8500 plasma is pretty shocking (0.02 fL and CR ratio under 2K:1)


I can't believe the guys at the HDTV Shootout overlooked something like that.


Does the Panasonic ST60/VT60/ZT60 have a similar issue?

 

[HDTVChallenged]

A bit off topic but,


Does anyone know the average "contrast ratio" a/o MLLs in a typical commercial movie theater? Bonus points for Film projection vs DLP. ... For reference purposes only.

 

[todd95008]



Quote:
Originally Posted by PlasmaPZ80U  /t/1415546/patterns-to-measure-mll-as-a-function-of-apl#post_23707402


that floating black level on the Samsung F8500 plasma is pretty shocking (0.02 fL and CR ratio under 2K:1)


I can't believe the guys at the HDTV Shootout overlooked something like that.


Does the Panasonic ST60/VT60/ZT60 have a similar issue?

Before I sold it, my 50ST60 had a constant black level of .003fL (or lower since this is the limit of my meter) no matter what APL was.

You can see above on Zoyd's plot (post 14) that the older Panasonic panels were like this too.

My F8500 is a 51 inch panel and I don't think they are as good as the larger versions (could be the physics of the smaller pixels on the 51" size) ??

Off topic but the worst part is with the smart apps and streaming content which for some reason shuts off the black optimizer.