or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Spears & Munsil HD Benchmark Blu-Ray 2nd Edition
New Posts  All Forums:Forum Nav:

Spears & Munsil HD Benchmark Blu-Ray 2nd Edition - Page 6

post #151 of 636
Quote:
Originally Posted by GeorgeAB View Post

Here is what I suggested for this test pattern:

"There could ideally be additional varying shadow detail in the dark part of the pattern to supplement the PLUGE provided. This would give the viewer additional steps and image components for peripheral judgement of preserving shadow detail with the biasing illumination in use."

In some cases, additional fine tweaking of the "brightness" (black level) setting may be desirable after the biasing illumination is implemented and adjusted. The presence of low level ambient light in a dark viewing environment has an affect upon the perception of black floor and subtle shadow detail. The human visual system can take as much as a half hour to fully adapt to dark room conditions. There are electro-chemical changes that occur in the retina over such a period of time. Professional mastering technicians are instructed to allow for this time to adapt to the tightly controlled viewing environment maintained in such work spaces before working on a video program. Such attention to detail can pay off for predominantly dark programs, such as the recent 'Lincoln' movie. This is also common in digital radiology reading rooms where subtleties of gray scale variations are critical when doctors examine x-rays of soft tissue. Perhaps this kind of information could be added to S&M's program notes on their web site to handle future inquiries about these two test patterns.

Best regards and beautiful pictures,
G. Alan Brown, President
CinemaQuest, Inc.
A Lion AV Consultants affiliate

"Advancing the art and science of electronic imaging"

thanks for explaining the purpose of those diagonal bars in the background of the pattern
post #152 of 636
I've measured the equal energy white patterns at ~30% energy load on my plasma with gamma @2.2 This is high enough to engage a moderate amount of limiting as peak white was reduced from 135 cd/m^2 to 115 cd/m^2. In my opinion this level is not very useful for plasma gamma calibration targeting on-screen levels of 5%-20% loading which are typical of a wide range of movies, and do not have significant limiting effects for normal calibration parameters.
Edited by zoyd - 5/5/13 at 3:52pm
post #153 of 636
Quote:
Originally Posted by zoyd View Post

I've measured the equal energy white patterns at ~30% energy load on my plasma with gamma @2.2 This is high enough to engage a moderate amount of limiting as peak white was reduced from 135 ftL to 115 ftL. In my opinion this level is not very useful for plasma gamma calibration targeting on-screen levels of 5%-20% loading which are typical of a wide range of movies, and do not have significant limiting effects for normal calibration parameters.

Zoyd,

So we're back to an 18% APL on a 5% window for grayscale?
post #154 of 636
I prefer mascior's 4% "constant" energy (actual loading varies between 12-16% on a 2.2 display). Results with the GCD 22% APL are almost identical though for gamma. For gray scale the pattern sizes/APL/energy loading make very little difference assuming you aren't doing something weird like using full windows.
post #155 of 636
I have not tried the equal energy patterns yet, so I can speak on em. Ive been playing with different sizes for the VT50, and so far I like the results of the 10% window.
post #156 of 636
Quote:
Originally Posted by zoyd View Post

I've measured the equal energy white patterns at ~30% energy load on my plasma with gamma @2.2 This is high enough to engage a moderate amount of limiting as peak white was reduced from 135 ftL to 115 ftL. In my opinion this level is not very useful for plasma gamma calibration targeting on-screen levels of 5%-20% loading which are typical of a wide range of movies, and do not have significant limiting effects for normal calibration parameters.

since you seem quite knowledgeable regarding this subject, if the S&M 2nd Edition disc uses 30% energy load, what about other commonly used pattern discs like the APL Large patterns on the AVSHD709 disc (Version 2D) and the 10% APL Windows on the GCD disc? what energy loads do they use?

also, how does one determine the energy load of the typical BD movie since factors like genre (horror films tend to be the darkest and animated films tend to be the brightest) and aspect ratio (black bars reduce APL and there are various sizes of black bars based on aspect ratios like 1.85:1, 2.39/2.40:1, and 2.35:1 not to mention no bars for 1.78:1)?

one more question is did you mean 135 cdm to 115 cdm?
post #157 of 636
yes, thanks for the unit catch, cd/m^2. I've measured a number of movies by setting the probe up to capture as much of the display as possible and average luminance measured this way is always well below 20% peak white.
post #158 of 636
Quote:
Originally Posted by zoyd View Post

yes, thanks for the unit catch, cd/m^2. I've measured a number of movies by setting the probe up to capture as much of the display as possible and average luminance measured this way is always well below 20% peak white.

what size APL windows best represent this value?
post #159 of 636
I have a question about evaluating motion resolution on my Samsung LED-LCD TV. I'd like to see what effect 'LED Motion Plus' has on moving picture resolution. Which patterns, in any, on this disc are best for that purpose?
post #160 of 636
Quote:
Which patterns, in any, on this disc are best for that purpose?

All of the patterns under Video Processing -> Motion.
post #161 of 636
Quote:
Originally Posted by PlasmaPZ80U View Post

what size APL windows best represent this value?

If you are talking about EE style pluge patterns like AVSHD and Mascior, then the 4% windows with 2% pluge boxes around the edges is a good fit to the ranges I've measured. The GCD disk is also a good fit using a different technique which fixes the average video level input.

Here is one such sampled movie with mean loading of 14.2% along with the loading ranges for some popular pattern sets and the new S&M set. This movie was sampled for 1 hour during predominantly daytime scenes under natural lighting so it's at the upper end as far as mean levels go of those I've sampled.

post #162 of 636
Quote:
Originally Posted by airgas1998 View Post

so i'm wanting to get started in diy calibration. all of the threads show primarily using the 709 disc or something similar. is there anything wrong with using this new version 2 or the older version in which I have. I thinking something like calman 5 with the i1dpro.

bump-anybody...
post #163 of 636
airgas,

It'd be a good idea to first find out what saturation percentage works best for your 8500.
post #164 of 636
Quote:
Originally Posted by zoyd View Post

I've measured the equal energy white patterns at ~30% energy load on my plasma with gamma @2.2 This is high enough to engage a moderate amount of limiting as peak white was reduced from 135 cd/m^2 to 115 cd/m^2. In my opinion this level is not very useful for plasma gamma calibration targeting on-screen levels of 5%-20% loading which are typical of a wide range of movies, and do not have significant limiting effects for normal calibration parameters.

I kinda sat this out for a few days while I tried to digest this remark, and after due consideration I have to respectfully disagree. For the purposes of setting gamma, the key factor is that all the windows have equal energy, not that the total energy is below a particular limit.

To return to first principles, and I realize you understand this, I'm just reiterating it for all the other folks reading this thread: the problem with normal window patterns is that the plasma panel produces less light than a strict gamma curve would imply when the display is under significant load. In normal material this isn't a huge problem because while the overall response may float from scene to scene, the intra-frame relative gamma is still preserved. The whole gamma curve, in essence, shifts up and down based on the overall energy in the image. However, the curve keeps the same overall power function. The shape of the curve within the scene doesn't change because the screen is under load.

And to return to window patterns, with regular windows, there is enough energy difference between the highest and lowest windows to cause the output curve to shift. The result is that you don't get a proper gamma curve by following the normal calibration procedure with standard windows. One good solution is to make the window patterns equal energy, by putting the same pixels on screen, arranged into a central window for measurement and various arrangements of the other levels around the edges. This is the approach we followed with the HD Benchmark, and we tested it with a variety of displays and measurement instruments before settling on a solution we were happy with.

The bottom line is that the equal energy windows on the HD Benchmark are, in fact, for gamma calibration on plasma and work very well for that purpose.
post #165 of 636
Don,

For plasmas with a strong ABL such as the Panasonic VT50 series, do you think that the equal energy windows are appropriate for grayscale adjustment? Seems that their APL doesn't reflect most content.
post #166 of 636
Quote:
Originally Posted by dmunsil View Post


The whole gamma curve, in essence, shifts up and down based on the overall energy in the image. However, the curve keeps the same overall power function. The shape of the curve within the scene doesn't change because the screen is under load.

And to return to window patterns, with regular windows, there is enough energy difference between the highest and lowest windows to cause the output curve to shift. The result is that you don't get a proper gamma curve by following the normal calibration procedure with standard windows. One good solution is to make the window patterns equal energy, by putting the same pixels on screen, arranged into a central window for measurement and various arrangements of the other levels around the edges. This is the approach we followed with the HD Benchmark, and we tested it with a variety of displays and measurement instruments before settling on a solution we were happy with.

The bottom line is that the equal energy windows on the HD Benchmark are, in fact, for gamma calibration on plasma and work very well for that purpose.

Unfortunately the shape of the transfer curve does change under load for plasmas and that's why it's important what load level is chosen and why I suggest a load level closer to real-world operating conditions. The transfer curve (level averaged and functional form) is most stable between energy loading values of between 5% and 20% of peak demand. Below 5% some plasma displays manipulate gamma during darker scenes, and above that you have the ABL. It is a happy coincidence that this range also happens to correspond to typical operating levels for a wide range of content.

However, there is a lot of room for subjectivity with gamma calibration so while my points are theoretically correct I do encourage people to give the patterns a try and see what they think. They should give you a transfer function that will boost highlight contrast somewhat during scenes with average luminance below the pattern level. This could be a beneficial effect for bright room display for example.
Edited by zoyd - 5/7/13 at 9:21am
post #167 of 636
Quote:
Originally Posted by dmunsil View Post

I kinda sat this out for a few days while I tried to digest this remark, and after due consideration I have to respectfully disagree. For the purposes of setting gamma, the key factor is that all the windows have equal energy, not that the total energy is below a particular limit.

To return to first principles, and I realize you understand this, I'm just reiterating it for all the other folks reading this thread: the problem with normal window patterns is that the plasma panel produces less light than a strict gamma curve would imply when the display is under significant load. In normal material this isn't a huge problem because while the overall response may float from scene to scene, the intra-frame relative gamma is still preserved. The whole gamma curve, in essence, shifts up and down based on the overall energy in the image. However, the curve keeps the same overall power function. The shape of the curve within the scene doesn't change because the screen is under load.

And to return to window patterns, with regular windows, there is enough energy difference between the highest and lowest windows to cause the output curve to shift. The result is that you don't get a proper gamma curve by following the normal calibration procedure with standard windows. One good solution is to make the window patterns equal energy, by putting the same pixels on screen, arranged into a central window for measurement and various arrangements of the other levels around the edges. This is the approach we followed with the HD Benchmark, and we tested it with a variety of displays and measurement instruments before settling on a solution we were happy with.

The bottom line is that the equal energy windows on the HD Benchmark are, in fact, for gamma calibration on plasma and work very well for that purpose.

so, you're basically saying keeping the APL fixed for all grayscale steps is far more important that what that fixed APL value actually is (in this case around 30%)?
post #168 of 636
I have a question about the RGB clipping vs. YCbCr clipping patterns. What does each show and what do the YCbCr ones show that the RGB ones don't? Prior to using this disc, I've only seen RGB clipping patterns so the YCbCr ones are quite new to me.
post #169 of 636
Quote:
What does each show and what do the YCbCr ones show that the RGB ones don't?

Which patterns are you asking about?

We have one pattern called clipping that servers a special purpose. In that pattern, the RGB boxes have YCbCr values that fall between 16-235 Y (16-240 CbCr) and only go above 235 when converted to RGB.

We have the contrast patterns in RGB and YCbCr that have boxes from 230-254.

We have the dynamic range high patterns in RGB and YCbCr. These are the same boxes from the contrast pattern, but combined so one pattern has YCbCr and the other has RGB so you can see the three channels at once.

YCbCr can produce RGB values >255. The RGB clipping boxes ensure that you get R, G, or B, that are in the range specified. The YCbCr pattern adjusts the channels you do not care about to ensure that the RGB values that come out don't go above 255. Clipping on either is not good, but clipping RGB is worse than clipping the YCbCr. Hope that makes sense.
post #170 of 636
Quote:
For plasmas with a strong ABL such as the Panasonic VT50 series, do you think that the equal energy windows are appropriate for grayscale adjustment?

During development they were tested on VT20, VT25, VT50, and lots of BT300s in studios. Joe Kane presented our EE patterns at a SMPTE meeting a couple of weeks ago. Sadly many were surprised to learn about ABL issues on plasma. They have all been using standard window patterns to calibrate the displays used to master movies.

Another issue we face is what gamma value to use. 2.4 is what all movies mastered on a BVM should be displayed at. When Plasma's started replacing BVMs, things changed a bit. Sadly the value used is all over the map. I know that HBO calibrates their displays using a 2.2 gamma. There is now an actual spec that defines the display gamma, which BT.709 failed to do. That is BT.1886.
post #171 of 636
Quote:
Originally Posted by zoyd View Post

Unfortunately the shape of the transfer curve does change under load for plasmas and that's why it's important what load level is chosen and why I suggest a load level closer to real-world operating conditions. However, there is a lot of room for subjectivity with gamma calibration so while my points are theoretically correct I do encourage people to give the patterns a try and see what they think. They should give you a transfer function that will boost highlight contrast somewhat during APL scenes with average luminance below the pattern level. This could be a beneficial effect for bright room display for example.

I can't find any reference to this shape change, and it doesn't jibe with my understanding of the plasma ABL circuitry. Which doesn't mean I'm right; we might have missed something. We were focused on the core issue of ABL making it impossible to use normal windows to measure gamma.

If there is a shape change at high APL, it's a much smaller issue than the overall issue of ABL causing big proportional brightness shifts when the picture level changes. That large issue is what EE windows, of any average picture level, are designed to rectify. So at minimum I'm confident that using EE windows will produce a better reading of gamma and grayscale than using any non-EE windows.

And, of course, there's a big tradeoff in that reducing the average picture level necessitates reducing the size of the window. As the window gets smaller, it gets harder to aim the meter and ensure that you're only measuring the window and not any of the dark area around it.
post #172 of 636
Quote:
Originally Posted by PlasmaPZ80U View Post

so, you're basically saying keeping the APL fixed for all grayscale steps is far more important that what that fixed APL value actually is (in this case around 30%)?

That's what I'm saying, yes.
post #173 of 636
Quote:
Originally Posted by PlasmaPZ80U View Post

I have a question about the RGB clipping vs. YCbCr clipping patterns. What does each show and what do the YCbCr ones show that the RGB ones don't? Prior to using this disc, I've only seen RGB clipping patterns so the YCbCr ones are quite new to me.

Just to add to what Stacey said, the bottom line is that when we have a clipping test for a single channel, where possible we ensure that the other channels are within the nominal range. That's not always possible, but with RGB it always is. For example, it's quite possible to generate an R value above 235 while keeping the Y, Cb, and Cr values in their nominal range (16-235 and 16-240). That allows you to see if R is being clipped separately from Y, Cb, or Cr.
post #174 of 636
Quote:
Originally Posted by JimP View Post

Don,

For plasmas with a strong ABL such as the Panasonic VT50 series, do you think that the equal energy windows are appropriate for grayscale adjustment? Seems that their APL doesn't reflect most content.

Most test patterns aren't similar to real content, by design. The whole point is to isolate a small issue and make it larger and easier to see. In some cases, it's important to make a particular pattern close in some aspect to real-world content, but that's not actually common.

The point of EE windows is to make all the windows have the same total energy on screen. Making the level similar to the average picture level of a real movie is not a primary design goal.
post #175 of 636
Quote:
Originally Posted by sspears View Post

Which patterns are you asking about?

We have one pattern called clipping that servers a special purpose. In that pattern, the RGB boxes have YCbCr values that fall between 16-235 Y (16-240 CbCr) and only go above 235 when converted to RGB.

We have the contrast patterns in RGB and YCbCr that have boxes from 230-254.

We have the dynamic range high patterns in RGB and YCbCr. These are the same boxes from the contrast pattern, but combined so one pattern has YCbCr and the other has RGB so you can see the three channels at once.

YCbCr can produce RGB values >255. The RGB clipping boxes ensure that you get R, G, or B, that are in the range specified. The YCbCr pattern adjusts the channels you do not care about to ensure that the RGB values that come out don't go above 255. Clipping on either is not good, but clipping RGB is worse than clipping the YCbCr. Hope that makes sense.

Quote:
Originally Posted by dmunsil View Post

Just to add to what Stacey said, the bottom line is that when we have a clipping test for a single channel, where possible we ensure that the other channels are within the nominal range. That's not always possible, but with RGB it always is. For example, it's quite possible to generate an R value above 235 while keeping the Y, Cb, and Cr values in their nominal range (16-235 and 16-240). That allows you to see if R is being clipped separately from Y, Cb, or Cr.

I was referring to the dynamic range high patterns in RGB and YCbCr that have boxes from 231-253. At my current contrast setting (90/100 on a Samsung LED-LCD), I can see all the bars on all six channels. If I raise contrast, green begins to clip before blue and red.

As a side note, how easily should I be able to see the highest box in each channel? Should it be closer to invisible or fully visible? If contrast 90 makes it easily visible and contrast 91 makes it invisible, is contrast 90 or 91 better?
Edited by PlasmaPZ80U - 5/8/13 at 10:00am
post #176 of 636
Quote:
Should it be closer to invisible or fully visible?

It will be difficult to see.
Quote:
If contrast 90 makes it easily visible and contrast 91 makes it invisible, is contrast 90 or 91 better?

If it were my display, I would set to 90. Sounds like a pretty coarse jump.
post #177 of 636
Man, with calibrators as well as Joe Kane presenting these patterns, this is getting interesting.
post #178 of 636
Quote:
Originally Posted by sspears View Post

It will be difficult to see.
If it were my display, I would set to 90. Sounds like a pretty coarse jump.

the help menu for dynamic range RGB states "when everything is correct, all of the boxes except the last should be visually discernible against the background, though the last three or four may be difficult to make out unless you view the pattern in a dark room with your eyes completely adjusted to the light"

this seems to suggest the opposite, which is confusing given your response above


(I know the real word impact of a one click difference in contrast would be next to nothing, but I want to know what the proper way to use this pattern is.)
post #179 of 636
however, the help menu for dynamic range YCbCr states "all of the boxes should be visible"
post #180 of 636
Quote:
Originally Posted by PlasmaPZ80U View Post

the help menu for dynamic range RGB states "when everything is correct, all of the boxes except the last should be visually discernible against the background, though the last three or four may be difficult to make out unless you view the pattern in a dark room with your eyes completely adjusted to the light"

this seems to suggest the opposite, which is confusing given your response above.

The last box is a different level than the background, and it's supposed to be reproduced at a different level on the screen. But on most displays, it's nearly or in some cases completely impossible to see the difference between it and the background, because it's only one level different.

The spacing of levels in video is intended to make a one-level difference not visually distinguishable, so smooth gradients will look smooth. This isn't always true, as anyone who has seen banding in video can attest.

The wording of the help text is so people don't go crazy trying to make that last level visible against the background. On most displays you can't make it visible no matter how you adjust Contrast, and that's OK. On your display, apparently it can be made quite visible, which suggests there's something quirky going on with the gamma curve at the high end, or maybe that you're extra sensitive to small level differences than the average individual. Given that, our preference would be to choose the setting that makes it visible. You could go the other way and it shouldn't cause big problems. But if you want our opinion, there it is. smile.gif
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Spears & Munsil HD Benchmark Blu-Ray 2nd Edition