Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Capturing and Editing Video > Avisynth Development

Reply
 
Thread Tools Search this Thread Display Modes
Old 20th June 2019, 08:20   #81  |  Link
nacho
Registered User
 
Join Date: May 2019
Posts: 5
Quote:
Originally Posted by videoh View Post
I released DGHDRtoSDR 1.12 (see first post in thread). Parameters have changed so please read the text file. The default values look much better with thecoreyburton's sample. Feedback will be appreciated on "best" default values. The Vaporsynth properties are now set correctly after the 2020->709 conversion.
Very cool just in time for me to do some conversions I want. I'm doing gladiator at the moment, playing around with the settings. Will post some screenshots with settings soon.
nacho is offline   Reply With Quote
Old 20th June 2019, 17:54   #82  |  Link
videoh
Registered User
 
Join Date: Jul 2014
Posts: 758
Thanks, nacho!
videoh is offline   Reply With Quote
Old 22nd June 2019, 13:07   #83  |  Link
thecoreyburton
Registered User
 
Join Date: Jul 2015
Posts: 109
Thanks for the update!

I took some time to play around with things and I think that now things look perhaps a little too blue, but having the options there to tweak is very useful. The contrast also looks much better.

I've spent the last few days doing a bit of research and did note something: the source is noted as using the "Display P3" color primaries. I played around using HDRTools, and whilst I'm still relatively new to HDR (and HDR > SDR), I did manage to get a reasonably good output and some (hopefully) useful comparisons.

Here are three reference scripts and also images produced with them:

1 - DGHDRtoSDR 1.2, base settings:
Code:
DGSource("test.dgi", FullDepth=True)
DGHDRtoSDR()
Spline64Resize(1920, 1080)



2 - HDRTools, basic settings:
Code:
DGSource("test.dgi", FullDepth=True)
ConvertBits(10)
Exposure=14
ConvertYUVtoXYZ(Color=0, OutputMode=1, HDRMode=0, FullRange=False)
ConvertXYZ_Reinhard_HDRtoSDR(Exposure_X=Exposure, Exposure_Y=Exposure, Exposure_Z=Exposure)
ConvertXYZtoYUV(pColor=2, Color=2, OutputMode=2)
Spline64Resize(1920, 1080)



3 - HDRTools, settings specified for "Display P3" (hopefully correct):
Code:
DGSource("test.dgi", FullDepth=True)
ConvertBits(10)
Exposure=14
ConvertYUVtoXYZ(Color=0, OutputMode=1, HDRMode=0, FullRange=False Rx=0.680, Ry=0.320, Gx=0.265, Gy=0.690, Bx=0.150, By=0.060)
ConvertXYZ_Reinhard_HDRtoSDR(Exposure_X=Exposure, Exposure_Y=Exposure, Exposure_Z=Exposure, pColor=0,  pRx=0.680, pRy=0.320, pGx=0.265, pGy=0.690, pBx=0.150, pBy=0.060)
ConvertXYZtoYUV(Color=2, pColor=0, pRx=0.680, pRy=0.320, pGx=0.265, pGy=0.690, pBx=0.150, pBy=0.060)
Spline64Resize(1920, 1080)



Note #1: Reinhard was chosen arbitrarily here and an exposure of 14 was what I felt was right, but both settings are subjective.
Note #2: I can provide the second source if needed, but it's currently not demuxed. The most significant difference in the properties here is that the master display luminance was higher (4000 vs 1000).

The two strongest things I noticed from these comparisons is that the blue level is much stronger in DGHDRtoSDR than it is in HDRTools' tonemapping. It's hard to say which is more accurate, but I'm leaning towards the warmer colors of the default HDRTools here.

Perhaps the most interesting thing here is the difference in colors between the two HDRTools tests. Despite the file being marked as "Display P3", ignoring those values gives a much more vivid blue sky and also enhances the red values of the dog's collar amongst other improvements. This, combined with comparison to the BD screenshot I provided previously, leads me to thing that perhaps the "Display P3" tag might have been something automatically set by the encoder, rather than something deliberately tagged. I could also be mistaken. The second source also uses "Display P3", but it's much harder to tell whether or not the tag represents the intended output colors on this source.

Is there any handling for different primaries and coefficients in DGHDRtoSDR? I'm wondering if this detection could have been hindering the results DGHDRtoSDR provided initially. If ignoring this somehow improves the situation, could something such as a manual override be added?

Last edited by thecoreyburton; 22nd June 2019 at 13:11.
thecoreyburton is offline   Reply With Quote
Old 23rd June 2019, 14:19   #84  |  Link
nacho
Registered User
 
Join Date: May 2019
Posts: 5
Hi @videoh do you have any tips regarding balancing the 'roll' and 'light' parameters. Both affect brightness and contrast correct? How should I go about testing correct values? Which should I adjust first - is there some methodology?

E.g. you can produce near-identical frames by lowering one and increasing the other: https://diff.pics/mym6TuC6fv5t

Should I just leave one and adjust the other? If so, which should I leave alone?

Last edited by nacho; 23rd June 2019 at 14:30.
nacho is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

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

Forum Jump


All times are GMT +1. The time now is 03:58.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.