help creating my own calibration patterns...

The discussion that follows http://www.avsforum.com/avs-vb/showt...2#post21562032 touches on some of the issues that visca blaugrana ran into with creating patterns. Basically my suggestion is to begin with something simple like recreating one of the example videos. If you decide that you want to make windows of a different size I attached the 10% gray images we use with the avs file for the 10% series, and you could always modify the area of the windows with an image editor if you wanted.

I see you've posted in the HCFR thread. If you hook your display to a computer and use the HCFR generated images you can have HCFR create windows of different sizes. Using HCFR would make it a lot easier to measure what happens with windows of various sizes with probably less time invested. Also the automatic measurements are usually quicker than manually changing patterns. Personally my computer can be set to match the other sources I own, so it's often quicker to use automatically generated patterns.

Whenever people get into window size discussions, what I primarily want to know is how studio monitors perform. Since it's rather clear that window measurements on consumer displays can be highly influenced by Average Picture Level, basically my interest is in knowing how studio monitors perform with both windows and various constant Average Picture Level measurements. Without this sort of information, I think it's often difficult to say if "gamma calibration" is actually moving closer to or farther from any intended target.

 

10gray.zip 100.634765625k . file

 

avs-file.zip 0.404296875k . file

I've also been interested in the effects of various pattern sizes/configurations on my Samsung display calibration and created some test patterns using the following (don't know if this workflow is something you can use though):

1. Create patterns in photoshop (0-255) and write to uncompressed tiffs
2. Import and transform to YUV in final cut pro (both SD and HD versions)

This seems to work quite well, I've benchmarked some 15% area windowed patterns against AVSHD and get great agreement in luminance values and dE.

I've also compared small embedded gray scale and color patterns within captured movie frames to simulate an APL based on real material. I analyzed a series of images from the movie "The Tree of Life" (which has wonderful natural looking cinematography) and found average video levels tend to be lower than I expected (25-30%) but you often get a bi-model histogram with means around 25% and a second smaller peak in the 60-70% range for typical sunlit scenes. So I tested a set of patterns using a test image with 27% average video level and embedded patterns at 55% (1 standard deviation above the mean). The results are summarized here:



Average comparison with AVSHD windowed, dE=0.6 +/- 0.6
Average repeatability, dE=0.7 +/- 0.4

The yellow highlights comparisons that yield >3 dE differences and red >6 dE. Using the 15% area windowed patterns as reference both the Scene based vs. windowed, and AVSHD Large APL vs. windowed showed >3 dE differences at the low end of the gray scale and in the primaries. The APL patterns showed the larger differences. Average gamma was quite stable for all the patterns on my contrast and gamma settings (2.29, 36 ftL peak white). I don't consider any of these differences dramatic, the largest measured difference of 7.1 dE is still quite hard to discern but it does show that at the 3-6 dE level the patterns you choose do make a difference on displays with APL dependent response.

I believe you're saying that you're averaging video levels. A 15% area white window would average to 15% video levels, and a 15% gray field would also average to 15% video levels. I think it's safe to say that a 15% area white window is much more difficult for Plasma or CRT to display than a 15% gray field. I'm not sure that there is a point in simply averaging video levels, since averaging video levels does not seem to relate with strain placed on the power supply of Plasma or CRT displays.

I think that when people generally talk about Average Picture Level they are referring to a gamma-adjusted level at the display. For example http://www.videoessentials.com/glossary.php has the quote:

"APL (Average Picture Level) The video signal level, during the active picture part of each horizontal line, is mathematically averaged over the period of a frame to come up with APL."

The link also lists the 15% APL for video that I've seen others use:

"Television program material is said to have a 15% APL over a long period of time."

"The computer world uses a much higher number for APL over time, that of 35%."

Using a gamma-adjusted average like the DVE link seems to be referring to, I think the Small APL patterns would be around 5% APL and the Large APL would be around 25% APL. At least that was the original intent, having the Small APL darker than average content and the Large APL as brighter than average content. For general use it probably would have made more sense to make one 15% APL series, but actually the APL patterns were created mainly to test how window measurements were not reflecting a change that was readily apparent when looking at grayscale bars, so I wanted two measurements.

yes I checked this, my plasma has no problems with average video levels up to 45% regardless of pattern configuration. A 15% area white window measures exactly the same as a 1% area white window. I'm not really interested in display performance with ABL so all the measurements are done well below the threshold where it becomes a problem. The only exception is the AVSHD large APL patterns, at my gamma/contrast ABL is just starting to kick-in so I would guess the average video level of those patterns is around 45%.

Attached is the 75% color series and the template used for the windows. If I was doing it, I'd probably use a square area for reduced-size windows. For example keeping the height and making the width the same would be around 8% area. I think you were asking for the saturation series, and depending if you plan to post your video for general use, I may be willing to post those at a later time.

 

75color.zip 87.3603515625k . file

I'm still not sure this is common plasma or CRT behavior from the measurements I've seen. When I did a simple average of the video levels for the Large APL pattern it's only 36%, which would seem to be be below your comment here. The way I read this statement 5%, 10%, 15% area windows and either the Small or Large APL would all measure similarly. That sort of behavior is not in line with what most plasma measurements seem to indicate.

Edit: My main focus with these comments is primarily Y, which would be the point of Bluescale's interest. The only way I might be able to make sense of the quote is if you were mainly commenting on xy, which seems to be the main focus of the previous chart.

Yes, I measure a constant white Y (last line in chart) for all patterns. I've separately measured where Y starts to dip and that's at 45% average video levels. Of course if I ran a higher contrast this would change but my peak white of 36 ftL is not overly dim. I wouldn't use CRT experience as a guideline to modern plasmas, I would guess the phosphors are much more efficient and there is a fundamental difference in how they are driven. Plasmas are constant voltage devices, light output is adjusted by the rate and width of the sustain pulse not the voltage. So it takes the same amount of current (energy) to drive 100 pixels at 50% as 50 pixels at 100%

Okay, that makes sense. Most of the plasma measurements I'm used to seeing people post don't operate that way.

My main reason for posting was to point out that in addition to ABL issues for which people recommend windowed areas of various sizes, there are differences (primarily in color response) unrelated to ABL. Although not large, some people do care about 3-6 dE differences and those can easily be obtained using small windows vs. the APL patterns for instance. The OP has a panasonic and I've measured more or less the same behavior on the GT30 and if using the mid or high panel brightness the shifts in color response between an idealized window pattern and real content is probably larger.

This whole thing is an exercise in futility.

Whatever window size you use will make different results. None of them will be "right" because no window pattern ever exists in video content. Even if you vary the window size (large windows for darker steps, smaller windows for brighter steps), you will STILL simply get a different result from constant size windows. Neither method is "best" or "ideal" or anything... they are just different.

18% window size (18% of screen area) was Joe Kane's choice for CRT displays. I find plasma results tend to be best with window sizes around 10%-12% of screen area. Anything else you do will simply be different... not better, not improved, not ideal. Plasma's brightness limiting makes any calibration attempt a "best compromise" situation. And since video content changes in dynamic ways you can't duplicate with patterns, ANY pattern you choose is just a compromise. All we can say for certain is that full-screen patterns are NOT the right choice for calibrating plasmas... how small you go... 18%, 12%, 10%, variable... there are few, if any, people who could tell the difference in plasma images with any of those window sizes after calibration. Each size will produce slightly different settings, but how visible those setting differences are in the calibrated images is over-rated.

I agree with Doug, I don't mean to imply any of the patterns are "correct". However I don't agree that providing and discussing the range of errors one encounters is futile. If anything it should help to tamp down some of the hyperbole associated with pattern choice.

Personally I have little interest in plasma, aside from it being a current reference, but do you realize that probably a majority of people have reported noticable differences between adjusting gamma using power-limited windows and constant average picture level patterns? Considering the number of people that seem to have an opinion one way or another on the idea of using constant average picture level patterns on plasma, I think they're generally talking about noticeable differences, rather than minutia. Sure Zoyd basically says his settings do not result in power-limiting, which would seem to agree with your view in the quote, but it's likely a majority of the plasma measurement comparisons I've looked at seem to show some amount of power-limiting, even using 10% windows.

I think we are talking about two different effects. It is true that I measure no ABL effects on my plasma when peak white is set at 36 ftL or below and input video levels do not exceed an average of 45%. I do however still measure differences in gamma response with windows vs. APL patterns. These differences are not caused by the ABL, they are caused by the DSP and/or drive systems which adjust gamma as a function of APL. While the table I posted indicates the average gamma is constant, if you look at the shape of the gamma response it can be calibrated flat(assuming you have 10pt controls which the GT30 does not) for either APL or window patterns but not both.

I've previously reported on this effect in both Panasonic and Samsung plasmas, there exists a "stable" gamma regime between roughly 20% and 50% average video level. So depending on what portion of this curve the patterns probe you will get different measured gamma response. Note also that the stable regime will shift depending on your own personal gamma and contrast settings.

The saturation patterns on the disc are just based on the ColorHCFR series. They were trying to maintain a constant light output with an assumed gamma as they changed saturation. It's discussed in the following thread http://www.avsforum.com/avs-vb/showt...9#post15619419

The patterns are attached.

 

sat.zip 344.8876953125k . file

I'm not sure how you're eliminating brightness limiting in the quoted portion. When you said "I measure a constant white Y" I took that to mean that you were not adjusting contrast depending on the series of patterns, and Y of white for all the patterns was equal. To me that would be different than adjusting contrast so that Y of white in each series had the same white level. I would tend to lump the second option into what I would call brightness limiting, where higher Average Picture Levels tend to cause white Y to drop.

That's correct, contrast is constant for all patterns. I don't know why you thought otherwise.

I was mainly trying to come up with some possible explanation how 100% stays relatively constant while lower levels fluctuate. Most posted measurements show a significant difference in Y of white between the Small APL and Large APL series, so I thought maybe you were making changes at the display depending on the patterns being measured to keep Y of white similar. Anyway, I still don't see what basis people are using when they do have access to gamma controls in trying to determine they are moving closer to a reference, when the reference is basically however the studio monitor performs.

That's the definition of APL dependent gamma. Peak white is maintained and all other levels are redistributed (Lower levels by a larger percentage compared to higher levels)


I haven't seen this but it would be the case if I maxed out my contrast. Again these are two different things, ABL and DSP/drive gamma shifting.


I would argue that targeting the stable gamma portion of the gamma vs. APL/AVL curve makes
the most sense.

Also when peak white is not the same, I'd say some amount of gamma variation is generally expected on the sorts of consumer displays in discussion, for example as shown on most of the comparisons between the Small and Large APL measurements.


I can't say I know exactly what this means, but I'll guess the quote might primarily relate back to post #19. I'm trying to make sense of how "roughly 20% and 50% average video level" is in agreement with Doug Blackburn that "plasma results tend to be best with window sizes around 10%-12% of screen area", since 70% gray on a 12% area window would only seem to be about 8% average video level. Again, my main focus is that people adjusting 10 point gamma controls seem to think that they're coming nearer to an intended target gamma, and I just don't see the basis behind such a thought.

I just know that their intent was to maintain a constant brightness if the display matched a target gamma. I think the point was they were trying to have only one variable, just a change in saturation. Since gamma measurements aren't exactly defined, the idea might work a little better in theory than actual practice, but saturation measurements can still be useful in some situations.

I don't understand the rational for 10-12% windows, maybe Doug can elaborate. It's far below any power limiting on 2011 plasmas and represents the APL of only the darkest scenes.

By "targeting stable gamma" I mean use a pattern which generates an APL in the range in which input levels are not being adjusted either on the high end (ABL) or low end (contrast enhancements). For example, with a peak white of 35 ftL and desired gamma of 2.2 this stable region is between 1 and 7 ftL. So I currently use an APL pattern with generates total light output of 2 ftL (27% input level). In contrast, a series of 10% windows generates an APL ranging from 0.001 ftL to 0.2 ft. That range of stimulus captures a tiny fraction of the operating space of the display. So for calibration purposes in my case I can say "My display gamma is 2.2 over an APL range of 1 to 7 ftL". This is the best one can do given the display's manipulation of input levels.