AVS APL Study - Adjunct to AVS Contrast Project - Page 5 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 23Likes
Reply
 
Thread Tools
post #121 of 178 Old 07-25-2015, 03:27 PM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
I thought I'd post this here since it's mainly about the avisynth code. When I saw Darin's post in the APL primer thread that North By Northwest was clocking in at 5% average luminance I was surprised based on some earlier measurements I had taken of some other movies. So I measured it taking a sample of a large area of the display once every 1.5 seconds while the movie was running. The value I came up with was 8% average luminance (using a 2.26 gamma display) so we've been trying to figure out what caused the difference.

I was not familiar with the avisynth script so I thought at first that it might not be calculating the proper colorimetric luminance. But after running some test patterns through the script and comparing results of still images where I can calculate the average luminance myself, the script does what it's supposed to do. I did however find two errors. The first is very serious if you are running test patterns but seems to have a small effect on the movie statistics. It turns out that the ConvertToRGB() step scales the video range input to full range so that when run through the Levels() conversion it gets scaled again, which is not correct. The other error was that the conversions between RGB and YUV default to SD colorspace using the Rec601 matrix, so for HD test clips you need to specify the Rec709 matrix. This has a negligible effect on the luminance statistics.

So I went ahead and updated the script and added a few new things:

1. Showchannels plugin [attached]

This displays a text overlay of real-time current frame and accumulated R'G'B' average luma (stimulus). These values are also saved to the output file and a separate summary file. So the output file will contain both average non-linear input luma and linear output luminance per frame sampled.

2. Changes to output file

Added a bunch of stuff like current time to see how long the processing took, frame time, color space flag, color range flag, %average luma for RGB channels, %average luminance. Also changed separator to a comma for .csv style output.

3. Fixed levels and colorspace conversion for video range HD input.

4. Added dithering switch to levels conversion and lanzcos chroma up-sample to ConvertToRGB().

The linearization step plus subsequent rounding back to 8 bits loses all information from input stimulus levels up to 6%, dithering will recover some of that info.

Note that the script requires both the FFVideoSource and ShowChannels[attached] plug-ins.

Sample output is also attached.

With this script North By Northwest results for average luminance were: Mean: 6.2% Median: 5.4% Max: 28.1%

Spoiler!
Attached Files
File Type: zip ShowChannels_25&26_v0-8_dll_20150330.zip (156.8 KB, 215 views)
File Type: zip nbnw.zip (429.3 KB, 35 views)

Last edited by zoyd; 09-25-2015 at 05:57 PM. Reason: correct showchannels overlay effects, recommend show=false
zoyd is offline  
Sponsored Links
Advertisement
 
post #122 of 178 Old 07-26-2015, 04:41 AM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Some results:

Cumulative average stimulus: RGB = 11.5% 15.0% 14.2%

Cumulative average stimulus: RGB = 14.5%, 14.5%, 15.4%

Cumulative average stimulus: RGB = 15.9%, 15.2%, 14.5%

Cumulative average stimulus: RGB = 13.7%

Cumulative average stimulus: RGB = 29.0% 24.4% 22.0%

Cumulative average stimulus: RGB = 19.4% 16.2% 15.7%

Cumulative average stimulus: RGB = 20.6%, 20.2%, 19.0%

Cumulative average stimulus: RGB = 20.7%

Cumulative average stimulus: RGB = 26.8% 25.0% 19.8% [R/B ratio -> oversaturated]

Cumulative average stimulus: RGB = 31.4% 27.2% 27.3%

Cumulative average stimulus: RGB = 25.4% 26.9% 25.1

Cumulative average stimulus: RGB = 42.4%

Last edited by zoyd; 05-23-2018 at 09:16 AM.
zoyd is offline  
post #123 of 178 Old 07-28-2015, 04:37 PM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by stanger89 View Post
These might be helpful for processing at greater than 10 bit
http://avisynth.nl/index.php/High_bi..._with_Avisynth
http://forum.doom9.org/showthread.php?t=170509
http://forum.doom9.org/showthread.php?t=154971

It sounds like AVISynth, while being limited to 8-bit can handle two stacked 8-bit clips effectively giving it 16 bit precision. I haven't investigated to understand how to make use of it. But SmoothAdjust has a SmoothLevels filter that has a gamma parameter.
For reference, here is a version of the script which will process the colorspace conversions and linearization in 16 bits with dithering. In addition to Showchannels, which will now report non-linear YUV, additional required plugins are:

1. Dither tools
2. Masktools 2

Also added automatic 709/601 and FPS processing.

I reran North By Northwest and there was no significant difference between the 16 bit dithered and 8 bit dithered results.

8 bit mean, median, max: 6.152%, 5.354%, 28.1%
16 bit mean, median, max: 6.157%, 5.420%, 27%

Spoiler!

Last edited by zoyd; 07-29-2015 at 04:49 AM.
zoyd is offline  
Sponsored Links
Advertisement
 
post #124 of 178 Old 08-02-2015, 08:44 AM
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,254
Mentioned: 190 Post(s)
Tagged: 0 Thread(s)
Quoted: 913 Post(s)
Liked: 1328
Hi Zoyd!

Thank you for your very big/helpful input!

I have a question:

the results you just showed above are with or without the black bars included? And with or without the intro/credits?

I'm trying to compare with the results I got with this script:

Quote:
LoadPlugin("H:\Video\Software\ffms2-2.20-icl\x86\ffms2.dll")
FFVideoSource("H:\Video\Casino_Royale\title00.mkv" , cachefile = "H:\Video\Casino_Royale\title00 temp files\title00.ffindex", fpsnum = 24000, fpsden = 1001)
Crop(0, 140, -0, -140)

filename = "CasinoRoyale"
ConvertToRGB()
Levels(16,1/2.2,235,0,255,coring=false)
ConvertToYV12()
Selectevery(24,0)
AssumeFps(240)
colon = " : "
WriteFile(filename, "current_frame", "colon", "YPlaneMin", "colon", "YPlaneMax", "colon", "AverageLuma" )
WriteFileStart(filename, """ "AverageLuma for CasinoRoyale" """)
WriteFileStart(filename, """ "Frame : MinLuma : MaxLuma : AverageLuma" """ )
WriteFileEnd(filename, """ "End" """)
ConvertToYUY2()
Can you tell me what the function "ConvertToRGB()" do when you do not fill the parenthesis with REC 701?

Are you sure the result is "wrong" with blu-ray sources? (except for your improvement with the dithering method to improve the results for the stimulus up to 6%):

For example with casino royale, I get (without black bars / with intro/credits):

Mean ADL: 11.8% (you find 11.1%)
Median ADL: 7,18% (you find 7.3%)


Seems close enough that the dithering alone you added could explain the differences?


Thank you!

Last edited by Soulnight; 08-02-2015 at 08:54 AM.
Soulnight is online now  
post #125 of 178 Old 08-10-2015, 11:08 AM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by Soulnight View Post
Hi Zoyd!

Thank you for your very big/helpful input!

I have a question:

the results you just showed above are with or without the black bars included? And with or without the intro/credits?
The intro and credits were included and the black bars removed.

Quote:
Can you tell me what the function "ConvertToRGB()" do when you do not fill the parenthesis with REC 701?
The default is to use Rec601 primaries if you don't specify Rec709.

Quote:
Are you sure the result is "wrong" with blu-ray sources? (except for your improvement with the dithering method to improve the results for the stimulus up to 6%):

For example with casino royale, I get (without black bars / with intro/credits):

Mean ADL: 11.8% (you find 11.1%)
Median ADL: 7,18% (you find 7.3%)


Seems close enough that the dithering alone you added could explain the differences?

Thank you!
Well to be consistent you should use the primaries that your display is using as that will best model the luminance it will produce. And if you are viewing blu-ray material your display should be calibrated to Rec709 primaries. Another factor that could account for the slightly different results is that I was using the real-time mean/max RGB display overlay and that actually changes the statistics a little bit, I updated the scripts to recommend that this feature should be turned off for the most accurate results.
zoyd is offline  
post #126 of 178 Old 08-11-2015, 10:02 PM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Thanks for your help here Zoyd. I was on vacation for some of this.
Quote:
Originally Posted by zoyd View Post
The linearization step plus subsequent rounding back to 8 bits loses all information from input stimulus levels up to 6%, dithering will recover some of that info.
This was my biggest concern, but I hadn't figured out how to do the calculations in 16 bit space. I was hoping that with some levels getting rounded up the error wouldn't be too high from this when looking for median image ADLs.

I need to figure out what I'm doing wrong as I tried to grab the right dither tools and put the dither.dll in an avisynth plugins directory, but get an error that dither_convert_yuv_to_rgb isn't found. I can look at it later.

The numbers I got that I'm most concerned about are for some material where I got lower medians than with North By Northwest. Using the 8 bit tools, 2.2 gamma, checking every frame, and using the scripts like before (probably not specifying REC.709) I got the following for median image ADLs after cropping and cutting off dark credits at the beginning and end:

Maleficent: 2.8%
The Hunger Games: M1: 2.3%

For the 10th episode of season 3 of Game of Thrones I got about 1% median. I recall trying both the DVD version and the HD version, but I would have to go check things.

I'm guessing you don't have any of those, but just in case you or anybody has them and wants to try one of them, I would be interested in results with the new scripts and 16 bit dithering. Hopefully I can get things work so I can get my own results with the new scripts.
Quote:
Originally Posted by zoyd View Post
I reran North By Northwest and there was no significant difference between the 16 bit dithered and 8 bit dithered results.

8 bit mean, median, max: 6.152%, 5.354%, 28.1%
16 bit mean, median, max: 6.157%, 5.420%, 27%
Were those with the BT.1886 gamma approximation? I wonder what you would get with 2.2 gamma. I would expect higher than with BT.1886.

Thanks,
Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."

Last edited by darinp2; 08-11-2015 at 10:05 PM.
darinp2 is offline  
post #127 of 178 Old 08-12-2015, 11:30 AM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by darinp2 View Post
I need to figure out what I'm doing wrong as I tried to grab the right dither tools and put the dither.dll in an avisynth plugins directory, but get an error that dither_convert_yuv_to_rgb isn't found. I can look at it later.
Make sure you put dither.avsi and mt_xxpand_multi.avsi in the plugins directory too.
Quote:
The numbers I got that I'm most concerned about are for some material where I got lower medians than with North By Northwest. Using the 8 bit tools, 2.2 gamma, checking every frame, and using the scripts like before (probably not specifying REC.709) I got the following for median image ADLs after cropping and cutting off dark credits at the beginning and end:

Maleficent: 2.8%
The Hunger Games: M1: 2.3%

For the 10th episode of season 3 of Game of Thrones I got about 1% median. I recall trying both the DVD version and the HD version, but I would have to go check things.
I'm guessing you don't have any of those, but just in case you or anybody has them and wants to try one of them, I would be interested in results with the new scripts and 16 bit dithering. Hopefully I can get things work so I can get my own results with the new scripts.
I can check the GoT episode.

Quote:
Were those with the BT.1886 gamma approximation? I wonder what you would get with 2.2 gamma. I would expect higher than with BT.1886.
The 8-bit dithered was using script #1 above (dither switch added to Levels()) with 2.2 gamma.
zoyd is offline  
post #128 of 178 Old 08-15-2015, 10:25 AM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Quote:
Originally Posted by zoyd View Post
Make sure you put dither.avsi and mt_xxpand_multi.avsi in the plugins directory too.
Thanks. That got me past that problem.


I'm not sure if I need any later versions of anything else, but I'm now getting this error:


Script error: There is no function name 'Dither_resize16'.
(C:/Program Files (x86)/AviSynth 2.5 plugins dither.avsi, line 826)
(C:/Program Files (x86)/AviSynth 2.5 plugins dither.avsi, line 1807)
(C:/Program Files (x86)/AviSynth 2.5 plugins dither.avsi, line 289)

that looks like it is from this line in the script:


FFCOLOR_SPACE == 1 ? dither_convert_yuv_to_rgb(matrix="709", mode=6, tv_range=true, output="rgb48y") : dither_convert_yuv_to_rgb(matrix="601", mode=6, output="rgb48y", tv_range=true)


Thanks,
Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #129 of 178 Old 08-15-2015, 11:18 AM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Not sure what the source of that problem might be, you can check dither.avsi and see that the function is in there. I'm using Avisynth 2.6 and corresponding masktools2 builds, you might try that.
zoyd is offline  
post #130 of 178 Old 08-15-2015, 12:34 PM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by zoyd View Post
I can check the GoT episode.
Results for GoT S03E10 (no borders, with credits) were: Mean=3.51%, Median=1.37%, Max=51.8%

Mean Y' = 12.58%

Last edited by zoyd; 05-23-2018 at 09:24 AM.
zoyd is offline  
post #131 of 178 Old 08-16-2015, 04:17 PM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Quote:
Originally Posted by zoyd View Post
Results for GoT S03E10 (no borders, with credits) were: Mean=3.51%, Median=1.37%, Max=51.8%

Mean Y' = 12.58%
Thanks Zoyd. Looking back at my results I got a median of 0.93% (which I put in my contrast ratio presentation as 1%) while you got 1.37% and I got a max of 59.8% while you got a max of 51.8%. Definitely not as close as I would like, but at least not out of the same ballpark, especially when considering whether some content is far from well represented by the ANSI CR test patterns. I figured that one had the most potential for error from the 8-bit limitation out of the things I analyzed.

Thanks,
Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #132 of 178 Old 08-17-2015, 09:06 AM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
Quote:
Originally Posted by darinp2 View Post
Looking back at my results I got a median of 0.93% (which I put in my contrast ratio presentation as 1%) while you got 1.37% and I got a max of 59.8% while you got a max of 51.8%.
I think the main error source in the original script was the double scaling and not the lack of dithering. Scaling to full levels twice will clip 0-15 and 236-255 on the second pass, and that would reduce the overall mean/median. If I run the same episode through the original script with the correct colorspace and levels the results are:

mean: 3.58%, median: 1.30%, max: 53.9%, mean Y' = 12.59%

which is not significantly different from the 16-bit dithered script.
zoyd is offline  
post #133 of 178 Old 08-17-2015, 11:54 AM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Quote:
Originally Posted by zoyd View Post
I think the main error source in the original script was the double scaling and not the lack of dithering. Scaling to full levels twice will clip 0-15 and 236-255 on the second pass, and that would reduce the overall mean/median. If I run the same episode through the original script with the correct colorspace and levels the results are:

mean: 3.58%, median: 1.30%, max: 53.9%, mean Y' = 12.59%

which is not significantly different from the 16-bit dithered script.
Thanks Zoyd. Just want to be clear. For these numbers I assume you mean from the original script you changed:

ConvertToRGB()
Levels(16,1/2.2,235,0,255,coring=false)
ConvertToYV12()

to:

convertToRGB(clip,"Rec709", chromaresample="lanczos")
clip = Levels(0, 1/2.2, 255, 0, 255, coring=false, dither=false)
ConvertToYV16(clip, false, "Rec709")

Is that right?

Thanks,
Darin
Ericglo likes this.

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #134 of 178 Old 08-17-2015, 06:41 PM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Quote:
Originally Posted by zoyd View Post
Not sure what the source of that problem might be, you can check dither.avsi and see that the function is in there. I'm using Avisynth 2.6 and corresponding masktools2 builds, you might try that.
In dither.avsi I see a function Dither_resize16nr, but not Dither_resize16, although Dither_resize16 is called in there like with src.Dither_resize16. My dither.avsi shows a date of 7/16/2015 at 9:47 am.


It is possible that I now have an issue with having both Avisynth 2.5 and Avisynth 2.6 on my machine as when I try the scripts with dithering and the Avisynth 2.5 moved to a new filename (so it won't be found in the path) I am back to the error of it not finding dither_convert_yuv_to_rgb even though I have those .avsi files in the Avisynth+ directory. If I put the Avisynth 2.5 directory back with its plugin directory containing just dither.avsi and mt_xxpand_multi.avsi or with other plugins in there also, then I get past the first error and again get the error that there is no Dither_resize16.


At least I am able to run the 8 bit version with the leveling changes.


Thanks,
Darin
Ericglo likes this.

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #135 of 178 Old 09-15-2015, 01:22 PM
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,254
Mentioned: 190 Post(s)
Tagged: 0 Thread(s)
Quoted: 913 Post(s)
Liked: 1328
Quote:
Originally Posted by zoyd View Post
I think the main error source in the original script was the double scaling and not the lack of dithering. Scaling to full levels twice will clip 0-15 and 236-255 on the second pass, and that would reduce the overall mean/median.

Thank you again for your help.

Could you explain more in details where did we double scale to full level in the old code and how was it bad? Thank you!
Soulnight is online now  
post #136 of 178 Old 09-15-2015, 01:55 PM
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,254
Mentioned: 190 Post(s)
Tagged: 0 Thread(s)
Quoted: 913 Post(s)
Liked: 1328
Quote:
Originally Posted by Soulnight View Post
Thank you again for your help.

Could you explain more in details where did we double scale to full level in the old code and how was it bad? Thank you!
Ok I think I know now...since you already explained it above!
Soulnight is online now  
post #137 of 178 Old 09-25-2015, 04:16 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Quote:
Originally Posted by zoyd View Post
8 bit mean, median, max: 6.152%, 5.354%, 28.1%
16 bit mean, median, max: 6.157%, 5.420%, 27%
Is this data from the .csv or the output file of showchannels? And which lines / parameter are you reporting?
Stereodude is online now  
post #138 of 178 Old 09-25-2015, 04:37 PM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
zoyd is offline  
post #139 of 178 Old 09-25-2015, 04:44 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Quote:
Originally Posted by zoyd View Post
The statistics are calculated from the last line of the csv file, average luminance for the sampled frame.
Got it. Thanks!
Stereodude is online now  
post #140 of 178 Old 09-25-2015, 05:19 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Pirates of the Caribbean: On Stranger Tides (2D):
16 bit mean, median, max: 3.348%, 1.581%, 45.529%

Pirates of the Caribbean: On Stranger Tides (3D Left Eye):
16 bit mean, median, max: 5.337%, 2.609%, 50.689%


Top and Bottom black bars were cropped, but credits were included.

Last edited by Stereodude; 09-26-2015 at 05:12 AM.
Stereodude is online now  
post #141 of 178 Old 09-25-2015, 05:36 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
@zoyd Your script in post 123 reports YUV data, the one in 121 reports RGB data. Is the YUV data preferred or more useful?
Stereodude is online now  
post #142 of 178 Old 09-25-2015, 05:56 PM
AVS Forum Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 6,804
Mentioned: 146 Post(s)
Tagged: 0 Thread(s)
Quoted: 1783 Post(s)
Liked: 1348
zoyd is offline  
post #143 of 178 Old 09-26-2015, 05:15 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Another movie, 2 different US BD releases.

Underworld: Awakening (2D):
16 bit mean, median, max: 3.323%, 1.863%, 82.553%

Underworld: Awakening 2D/3D single disc (2D / Left Eye):
16 bit mean, median, max: 3.389%, 1.853%, 58.071%


The numbers here are somewhat misleading. The mean and median are nearly identical, but a quick visual check of the two makes it quite clear that they're not the same. There is definitely a different color cast to the 2D (LE) version of the 2D/3D disc from the standard 2D release. It also has slightly elevated black levels and reduced brightness in highlights. However, the gamma of the two does not seem very different. I have screenshots in this thread where I'm investigating differences in 2D Blu-ray and 2D/3D Blu-ray releases.
Stereodude is online now  
post #144 of 178 Old 09-26-2015, 05:15 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Quote:
Originally Posted by zoyd View Post
There's no particular benefit - those were the formats available from the two different tool sets.
Alright, thanks for all the help.
Stereodude is online now  
post #145 of 178 Old 09-26-2015, 09:14 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Another movie, 2 different US BD releases.

Resident Evil: Afterlife (2D):
16 bit mean, median, max: 14.750%, 5.960%, 98.294%

Resident Evil: Afterlife 2D/3D single disc (2D / Left Eye):
16 bit mean, median, max: 14.763%, 5.988%, 98.214%

Black bars were cropped, credits were kept. Visual comparison couldn't find any differences in brightness, color, contrast, or gamma.
Stereodude is online now  
post #146 of 178 Old 09-26-2015, 12:00 PM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Thanks for adding these new ones Stereodude.

Do you happen to have Guardians in the Galaxy? I used the left eye from the 3D version for that one, but not sure how close it is to the 2D version.

Thanks,
Darin
darinp2 is offline  
post #147 of 178 Old 09-26-2015, 12:08 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Quote:
Originally Posted by darinp2 View Post
Do you happen to have Guardians in the Galaxy? I used the left eye from the 3D version for that one, but not sure how close it is to the 2D version.
Yes, I have it. I can run it. I was starting with movies where I had a combo 2D/3D single disc and a separate 2D only release to see if the 2D releases were or weren't the same thing.

I was less than pleased that my 64" Plasma (Samsung F8500) was juicing the gamma in 3D mode (~1.55) and the "3D only" discs were also juiced. Near black just glowed. User menu adjustments of the set could only get the gamma up to ~1.8. Zoyd helped me with the service menu and I ended up with a gamma of ~2.05 or so with 10-50 at 2.15 or so with it tapering down above a bit that and a fairly large drop at 90%. 3D seems like the wild, wild west with no strict standards of what the display, player, and discs should or shouldn't do. Only one of them, if any, should be juicing the gamma, and certainly not more than one.

Last edited by Stereodude; 09-26-2015 at 12:11 PM.
Stereodude is online now  
post #148 of 178 Old 09-26-2015, 04:50 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Quote:
Originally Posted by darinp2 View Post
Do you happen to have Guardians in the Galaxy? I used the left eye from the 3D version for that one, but not sure how close it is to the 2D version.
Here you go.

Two different US BD releases.

Guardians of the Galaxy (2D):
16 bit mean, median, max: 5.359%, 2.162%, 85.799%

Guardians of the Galaxy "3D only" disc (Left Eye):
16 bit mean, median, max: 7.325%, 3.237%, 84.792%

Black bars were cropped, credits were kept.
Stereodude is online now  
post #149 of 178 Old 09-26-2015, 04:54 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,101
Mentioned: 36 Post(s)
Tagged: 0 Thread(s)
Quoted: 4113 Post(s)
Liked: 2930
Another movie, two different US BD releases.

Avatar (2D):
16 bit mean, median, max: 9.070%, 7.066%, 99.171%

Avatar 2D/3D single disc (2D / Left Eye):
16 bit mean, median, max: 9.151%, 7.141%, 99.189%

Credits were included, no cropping because it's 16:9.

From the numbers and a visual comparison I'd say they're the same with no gamma, brightness, contrast, or color adjustments. I expect the slight differences in the percentages are due to inherent AVC/MVC codec non-determinism.
Stereodude is online now  
post #150 of 178 Old 09-26-2015, 06:23 PM
AVS Forum Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 23,188
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1135 Post(s)
Liked: 1767
Quote:
Originally Posted by Stereodude View Post
Here you go.

Two different US BD releases.

Guardians of the Galaxy (2D):
16 bit mean, median, max: 5.359%, 2.162%, 85.799%

Guardians of the Galaxy "3D only" disc (Left Eye):
16 bit mean, median, max: 7.325%, 3.237%, 84.792%
Thanks. I never got the 16 bit stuff to work, but using the 8 bit version I got a median around the same for the 3D version. Interesting that the 2D version is so much less. If I had the 2D version it looks like it would have had the lowest median APL of the top 5 worldwide grossing movies of 2014 that I compiled values for.

While you are engaged in this I would appreciate it if you could help with a sanity check, just to get more confidence that we are analyzing these correctly.

I cut a 1 second clip out of North By Northwest from the brightest part I found and put it here:

https://www.dropbox.com/s/lub5hchysa...t1sec.avi?dl=0

When I ran the data for every frame of the movie I got about 30% ADL.

I extracted a bitmap of one frame, although it may have been expanded to PC levels:

https://www.dropbox.com/s/yo1kwozcd8...Frame.bmp?dl=0

I posted a couple of pictures of my screen with the waveform monitor on a Panasonic AE8000 with the same frame:

https://www.dropbox.com/s/1c9nnc8xji..._6062.JPG?dl=0

https://www.dropbox.com/s/5qw55glrur..._6061.JPG?dl=0

I know this isn't perfect, but just guesstimating, if the average was about 60% video level then with 2.2 gamma the ADL should be a little over 30%, although it if leans more toward blue that would take the ADL down. My memory is that when I had the AE8000 and a meter I measured closer to the upper 30s for the amount of light getting into my room off the screen, but I could be remembering incorrectly.

I can look later to see what I got for each frame in this small section when I ran the whole move, or run it again with just the small segment.

Thanks,
Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
Sponsored Links
Advertisement
 
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off