Improving Madvr HDR to SDR mapping for projector - Page 197 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 4754Likes
Reply
 
Thread Tools
post #5881 of 6033 Old 05-13-2019, 12:37 PM
Advanced Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 835
Mentioned: 134 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 685
@madshi I tested different playback resolutions, but it doesn't change anything. I'm testing @1080p .

The output gamut I use is BT.709, but changing the output gamut doesn't change anything either.

However, I always test with D3D11 native, and I get different results by switching to D3D11 copy-back or other hardware decoders :

Code:
Fer15          : Atomic Blonde 6124 -> 28,58 10,88 11,79 16,22 10,88 10,88 11,79 16,22
D3D11 native   : Atomic Blonde 6124 -> 28,61 12,14 12,72 17,91 12,14 12,14 12,72 17,91
D3D11 copy-back: Atomic Blonde 6124 -> 28.59 10.91 11.84 16.28 10.91 10.90 11.84 16.28
DXVA2 native   : Atomic Blonde 6124 -> 27.76 10.93 11.87 16.38 10.93 10.93 11.87 16.38
DXVA2 copy-back: Atomic Blonde 6124 -> 28.59 10.91 11.84 16.29 10.91 10.91 11.84 16.29
None           : Atomic Blonde 6124 -> 28.59 10.91 11.84 16.29 10.91 10.91 11.84 16.29
But it should not matters, unless we merge results.

Fer15 doc doesn't have much false positives though, I think that can be the reason why we don't get the same best algo.

When I begin testing, I always get fluctuating best algo until a certain amount of cases, then it stabilizes.

After checking the scenes from Fer15's doc and adding them to mine, I still get the same winner:




Updated doc: https://www.mediafire.com/file/4wq78..._V11.xlsx/file


Edit: Thanks to @Fer15 for sending me the samples of false positives, now there are even more!




Last version: https://www.mediafire.com/file/dpwkz..._V12.xlsx/file

I guess that's it for me
stevenjw, madshi, Manni01 and 4 others like this.

Last edited by Neo-XP; 05-13-2019 at 02:06 PM.
Neo-XP is offline  
Sponsored Links
Advertisement
 
post #5882 of 6033 Old 05-13-2019, 03:05 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,571
Mentioned: 456 Post(s)
Tagged: 0 Thread(s)
Quoted: 2330 Post(s)
Liked: 3002
I'm surprised that "D3D11 native" seems to be the odd one out. I didn't expect that. Maybe my rendering path is different there, for some reason, I'll have to double check. IIRC Fer15 uses software decoding. Probably that explains the difference between your measurements?
madshi is offline  
post #5883 of 6033 Old 05-13-2019, 06:39 PM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 606
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 298
Madshi, is there any hope for black bar detection being available with D3D11 native? I use DXVA2, as almost everyone in my black-bar-detecting situation does. Just wanted to volunteer my testing services (it can't all fall on Neo and Fer!) if that is something you are going to work on at some point (if it is even possible---it may not be!).
ddgdl is online now  
Sponsored Links
Advertisement
 
post #5884 of 6033 Old 05-14-2019, 12:45 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Quote:
Originally Posted by ddgdl View Post
Madshi, is there any hope for black bar detection being available with D3D11 native? I use DXVA2, as almost everyone in my black-bar-detecting situation does. Just wanted to volunteer my testing services (it can't all fall on Neo and Fer!) if that is something you are going to work on at some point (if it is even possible---it may not be!).
You can use D3D11 copyback, it works very well with black bars detection.

It would be nice if D3D11 native could support black bars detection at some point, because of the better performance. I have a lot more headroom with native than with copyback, and also we wouldn't have to move back and forth between native for measuring files and copyback for software features.

Unfortunately jRiver also needs copyback for UHD BD menus, so unless this can be supported with native too, it looks like copyback is here to stay for jRiver users who want menus...
stevenjw likes this.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is online now  
post #5885 of 6033 Old 05-14-2019, 01:43 AM
Member
 
Join Date: Sep 2018
Posts: 47
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 27 Post(s)
Liked: 11
same problem as you : d3d11native is quicker for madmeasure but i need to use d3d11copyback to detect blacks bars


RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceD3D11")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceD3D11Desc")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccel")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceDXVA2")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceDXVA2Desc")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelCUVIDXVA")
contains the differents values to script the automatic modification



for example :
Regwrite("HKEY_CURRENT_USER\Software\LAV\Video\HWA ccel","HWAccel","REG_DWORD",5)

means ;5=d3d11 3=dxva2 copyback 4=dxva2 native



read and save the value for differents configurations and restore the config you need
xxxx5 is offline  
post #5886 of 6033 Old 05-14-2019, 02:11 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Quote:
Originally Posted by xxxx5 View Post
same problem as you : d3d11native is quicker for madmeasure but i need to use d3d11copyback to detect blacks bars


RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceD3D11")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceD3D11Desc")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccel")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceDXVA2")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelDeviceDXVA2Desc")
RegRead("HKEY_CURRENT_USER\Software\LAV\Video\HWAc cel","HWAccelCUVIDXVA")
contains the differents values to script the automatic modification



for example :
Regwrite("HKEY_CURRENT_USER\Software\LAV\Video\HWA ccel","HWAccel","REG_DWORD",5)

means ;5=d3d11 3=dxva2 copyback 4=dxva2 native



read and save the value for differents configurations and restore the config you need
Thanks, my batch file already does this automatically with reg files loading, but I might use your commands instead in the next version as it seems to be path independent and allows to change only the specific key

I need to upload a new version with automatic measurements of BDMV folders, now that @Soulnight has updated their tool, it works great and is fully automatic, at least for MPC-BE and jRiver users. MPC-HC is a bit hit and miss. EDIT: what is the value for D3D11 Copyback if 5 is native?

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-14-2019 at 02:16 AM.
Manni01 is online now  
post #5887 of 6033 Old 05-14-2019, 04:09 AM
Member
 
Join Date: Sep 2018
Posts: 47
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 27 Post(s)
Liked: 11
@Manni01

EDIT: what is the value for D3D11 Copyback if 5 is native?
HWA ccel =5 means d3d11


HWAccelDeviceD3D11","REG_DWORD",0xffffffff
AND
"HWAccelDeviceD3D11Desc","REG_DWORD",0
means native

while

HWAccelDeviceD3D11","REG_DWORD",0
AND
"HWAccelDeviceD3D11Desc","REG_DWORD",7170
means copyback

7170 is for my nvidia, not sure if it's the same value for another card
Manni01 likes this.
xxxx5 is offline  
post #5888 of 6033 Old 05-14-2019, 04:33 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Great thanks!

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is online now  
post #5889 of 6033 Old 05-15-2019, 04:30 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
@Everyone

I release here the V4.0 of my batch file to run measurements files. In addition to the already supported extensions, it now supports measuring the main playlists (movies) and all the playlists (TV Series) for UHD BD Folders automatically, thanks to @3ll3d00d 's excellent utility, madMeasurer. [EDIT: the batch file doesn't support physical discs or ISOs as it is]

[EDIT: V4.01 has Native and Copyback set to OFF by default so that no registry change takes place unless you actively set these to ON].
[EDIT: V4.02 has a missing path for madmeasurer restored].

As @Soulnight has kindly added path support for already measured .mpls files in their utility (thanks again Anna and Flo!) the combo batch file V4.0 + Soulnight's utility works well for UHD BD Folders users. I'll give a bit more feedback in the other thread as soon as I can.

This new version also optionally sets D3D11 to native / restores it to copyback using direct DOS commands inspired from the autoit commands that @xxxx5 provided above. This makes it possible to switch decoding mode without using saved registry files, hence without risking overwriting other LAV video settings.

NOTE: Please don't post about this batch file in this thread or contact my by PM about it. Consider it an unsupported template. If you have any questions, please read the notes as well as the documentation of the tools used, and contact their authors in the respective thread/forum to discuss these.

Many thanks to all those who have made these improvements possible, and of course to you Madshi for all your ongoing work on HDR tonemapping. I can't believe how well the live algo performs in this latest build and it's scary to think there is still room for progress...
Attached Files
File Type: zip MadMeasureHDR Batch File by Manni V4.02 16-05-19.zip (2.9 KB, 23 views)

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-16-2019 at 04:02 PM.
Manni01 is online now  
post #5890 of 6033 Old 05-15-2019, 06:53 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,838
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6781 Post(s)
Liked: 6395
Quote:
Originally Posted by Manni01 View Post
for my current 85nits actual peak.
Is that with your RS2000? That seems really low? I get 85 nits in low lamp on my X9500 on a 120" screen?

Or are you stopping down the iris heaps. Just stood out as odd since I have never seen you say you had anything lower than 100 nits at least.

I just went from 85 to 112 in high lamp on my 9500 recently and am enjoying the slight uptick in highlight punch.
mkohman likes this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #5891 of 6033 Old 05-15-2019, 11:32 PM
Advanced Member
 
mkohman's Avatar
 
Join Date: Jun 2009
Location: London, United Kingdom
Posts: 769
Mentioned: 20 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 340
Quote:
Originally Posted by Javs View Post
Is that with your RS2000? That seems really low? I get 85 nits in low lamp on my X9500 on a 120" screen?

Or are you stopping down the iris heaps. Just stood out as odd since I have never seen you say you had anything lower than 100 nits at least.

I just went from 85 to 112 in high lamp on my 9500 recently and am enjoying the slight uptick in highlight punch.

Hi Javs, hope your well. I have the 9900 and I project from approx. 5.5m away from lens to screen. My screen is 135" 2.39 and I use the zoom function. I measured my nits whilst zoomed to 2.39 on low lamp and it is around 52 nits. on High lamp it is around 72 nits. Does this sound normal? That is with the iris fully open. Currently I have it clamped down to -10 on low lamp and I am happy with the picture its just perfect. I have dark walls and ceiling so I am pretty sure that helps too. If I fully open the lens I get 52 nits on low lamp but picture is sometimes brighter than I would prefer. Can you please advise if I am doing it right or shall I have the iris fully opened for HDR? thank you.
mkohman is offline  
post #5892 of 6033 Old 05-15-2019, 11:41 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Quote:
Originally Posted by Javs View Post
Is that with your RS2000? That seems really low? I get 85 nits in low lamp on my X9500 on a 120" screen?

Or are you stopping down the iris heaps. Just stood out as odd since I have never seen you say you had anything lower than 100 nits at least.

I just went from 85 to 112 in high lamp on my 9500 recently and am enjoying the slight uptick in highlight punch.
Off topic but yes I'm experimenting at the moment with the iris closed to -8 due to the DI issues on the rs2000. -8 is the sweet point here to get rid of 90% of the DI artifacts (both the yellowing and the blooming) while still getting as much brightness as possible. This is in low lamp because I don't want to use high lamp, especially in the summer (higher fan noise, a lot more heat, more power use, shorter bulb life). So it's 85nits in low lamp manual iris at -8 with the DI for now.

I haven't decided to stay there, but with the right settings the brightness is fairly comparable (in fact with Soulnight's new utility it's often too bright due to the new algo, I have to make some adjustments to dynamic tuning). It's mostly the highlights as you say that are a bit more compressed, but the better black floor/fade to black and the higher native on/off (40,000:1 vs 30,000:1) make it an interesting compromise, especially in my bat loft.

Still experimenting. I might go back to no DI and iris open in low lamp which gives me around 120nits. If I do, it won't be to get a brighter picture but to get 1:1 tonemapping (no highlights compression) more of the time. What I'd really like is 150nits peak with a great black floor/fade to black and no DI artifacts, but it looks like I'll need to wait for a laser model to get that. I can get this in high lamp, iris open with the rs2000 but without a DI and the fan noise, heat power use and raised black floor is more than I can handle. I still love my rs2000 though
mkohman and Javs like this.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-15-2019 at 11:54 PM.
Manni01 is online now  
post #5893 of 6033 Old 05-16-2019, 03:02 AM
Advanced Member
 
Rathbone's Avatar
 
Join Date: Nov 2006
Location: Germany
Posts: 978
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 56 Post(s)
Liked: 25
Quote:
Originally Posted by Manni01 View Post
Thanks, my batch file already does this automatically with reg files loading, but I might use your commands instead in the next version as it seems to be path independent and allows to change only the specific key

I need to upload a new version with automatic measurements of BDMV folders, now that @Soulnight has updated their tool, it works great and is fully automatic, at least for MPC-BE and jRiver users. MPC-HC is a bit hit and miss. EDIT: what is the value for D3D11 Copyback if 5 is native?
Hello Manni,

does your batch tool also process BDMV folder from disc? When I run it I get the message that the measurement file cannot be created (cause it tries to write it into the disc BDMV path).
Rathbone is offline  
post #5894 of 6033 Old 05-16-2019, 03:23 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Quote:
Originally Posted by Rathbone View Post
Hello Manni,

does your batch tool also process BDMV folder from disc? When I run it I get the message that the measurement file cannot be created (cause it tries to write it into the disc BDMV path).
Not as it is because I don't need it, I rip all my discs to BD Folders on network shares, but there might be options in @3ll3d00d 's tool to do this. You'll have to look at the instructions on Github (link in the batch file release notes).

You could also take a look at @xxxx5 's utility that supports ISOs, it didn't work for me from a batch file for BD folders but it seemed to be working well with ISOs. See the discussion from here: https://www.avsforum.com/forum/24-di...l#post57810196. There is a separate thread for this tool mentioned in the discussion, it would be a better place to discuss this than this thread as I don't want to clutter it while Madshi is doing this work.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-16-2019 at 03:26 AM.
Manni01 is online now  
post #5895 of 6033 Old 05-16-2019, 05:49 AM
AVS Forum Special Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 3,786
Mentioned: 235 Post(s)
Tagged: 0 Thread(s)
Quoted: 2267 Post(s)
Liked: 2308
My tool supports ISO, it will mount it and write into the iso then unmount it. No support though for writing to a separate directory.
Manni01 likes this.
3ll3d00d is online now  
post #5896 of 6033 Old 05-16-2019, 07:51 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
(split from earlier post after posting V4.02 of my batch file)

@madshi

I haven't had the time to do much detailed testing related to the new algos, but I like what I see with the settings attached, adapted from Neo-XP's settings for my current 85nits actual peak. Thanks a lot for the improvements. @Neo-XP and @Fer15 are doing such a great job that there is no need to interfere... Thanks to both of them for all the detailed feedback they tirelessly provide on each test build. A big thumbs up for all their work

I have two minor requests to suggest for your to do list if they aren't there already, whenever appropriate and if you see fit, as I know you have other priorities currently:

- Please could you fix the dynamic clipping option so that we can use it in the live algo, or explain why we can't use the setting we use with measurements files when using Soulnight's utility? The default value needs to be lowered significantly otherwise the option if enabled crushes highlights very badly. Currently I disable it for testing and for normal use.
- As Nevcairiel has done the necessary changes in LAV/jriver, please could you support measurements files when playing a bd folder from the menu? Right now playing a title with menus from jRiver means no measurements files. They are supported fine if playing the main movie only.

Not a request but a question: can you share with us when you plan to work on the "looking ahead" part of the live algo? Is this after the next two builds? Later? Just curious, of course don't reply if you prefer not to say. I'm also looking forward to seeing ported in the live algo the improvements brought by Anna and Flo in their utility regarding sky detection. It also looks very promising!
SOWK likes this.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-16-2019 at 08:27 AM.
Manni01 is online now  
post #5897 of 6033 Old 05-16-2019, 04:01 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
I've done more tests tonight at around 100nits peak and I like the settings attached better. More depth in the picture and a lot more punch. Possibly too bright to my eyes.

Using dynamic clipping with a value of 10 max brightens the picture and increases contrast without hurting the highlights too much.

I need to do more tests but I have very little time at the moment, sorry.
Attached Thumbnails
Click image for larger version

Name:	Settings Manni Build 79 16-05-19.JPG
Views:	79
Size:	189.8 KB
ID:	2567718  
tommarra likes this.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is online now  
post #5898 of 6033 Old 05-16-2019, 04:04 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,838
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6781 Post(s)
Liked: 6395
Quote:
Originally Posted by Manni01 View Post
I've done more tests tonight at around 100nits peak and I like the settings attached better. More depth in the picture and a lot more punch. Possibly too bright to my eyes.

Using dynamic clipping with a value of 10 max brightens the picture and increases contrast without hurting the highlights too much.

I need to do more tests but I have very little time at the moment, sorry.
I can confirm that the dynamic clipping setting of around 15 is quite good and not really detrimental, I spent about an hour testing just that setting a couple weeks back with a LOT of shots to try and find the point where it would help, but not kill highlights in some shots. It seems to be right around that number.
Manni01 and tommarra like this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #5899 of 6033 Old 05-16-2019, 08:44 PM
Advanced Member
 
tommarra's Avatar
 
Join Date: Aug 2013
Location: USA
Posts: 561
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 570 Post(s)
Liked: 337
Quote:
Originally Posted by Manni01 View Post
I've done more tests tonight at around 100nits peak and I like the settings attached better. More depth in the picture and a lot more punch. Possibly too bright to my eyes.



Using dynamic clipping with a value of 10 max brightens the picture and increases contrast without hurting the highlights too much.



I need to do more tests but I have very little time at the moment, sorry.


Manni - what would you recommend for 80 nits?


Sent from my iPhone using Tapatalk

Gear: Fronts: B&W 803D3 | Center: B&W HTM2D3 | Surrounds: B&W 705S2 | Backs: B&W DM601S3 | Heights: Definitive Techonlogy ProCinema 1000 | Receiver & Amps: Anthem MRX 720 + Emotiva BasX-5 | Projector: JVC NX7 | Source: HTPC with Nvidia 2070, Kodi + External Player: MPC-HC with MadVR (for Blu-ray rips), Nvidia Shield (for streaming content)
tommarra is online now  
post #5900 of 6033 Old 05-16-2019, 08:46 PM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 606
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 298
Quote:
Originally Posted by tommarra View Post
Quote:
Originally Posted by Manni01 View Post
I've done more tests tonight at around 100nits peak and I like the settings attached better. More depth in the picture and a lot more punch. Possibly too bright to my eyes.



Using dynamic clipping with a value of 10 max brightens the picture and increases contrast without hurting the highlights too much.



I need to do more tests but I have very little time at the moment, sorry.


Manni - what would you recommend for 80 nits?


Sent from my iPhone using Tapatalk
And remember, those are the settings without using measurement files. With measurement files I don't think you need any of those, right?
ddgdl is online now  
post #5901 of 6033 Old 05-16-2019, 11:47 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
Quote:
Originally Posted by tommarra View Post
Manni - what would you recommend for 80 nits?
The same (though they need finetuning), they are a bit too bright to my eye. Best is to decide on the settings yourself, it also depends on room, taste, calibration accuracy, display or projector used, etc. If the picture is too bright with these even with 80 nits, raise dynamic tuning in increments of 5 or 10. If too dim, lower dynamic tuning. You can put your real peak as well, but it shouldn’t make much of a difference.

Quote:
Originally Posted by ddgdl View Post
And remember, those are the settings without using measurement files. With measurement files I don't think you need any of those, right?
Well, yes, Madshi is asking for feedback on the live algo at the moment in this thread, not on measurements files, so the files are disabled with these settings.

I’ll post my settings with measurements files in the other thread, when I find the time to do more testing with Soulnight’s new algo.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders

Last edited by Manni01; 05-16-2019 at 11:53 PM.
Manni01 is online now  
post #5902 of 6033 Old 05-17-2019, 05:17 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 606
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 298
Quote:
Originally Posted by Manni01 View Post
Well, yes, Madshi is asking for feedback on the live algo at the moment in this thread, not on measurements files, so the files are disabled with these settings.

I’ll post my settings with measurements files in the other thread, when I find the time to do more testing with Soulnight’s new algo.
That was more for him than you- I was pretty sure he was using measurement files (though I could be confusing him with someone else), and was just flagging that.
Manni01 likes this.
ddgdl is online now  
post #5903 of 6033 Old 05-17-2019, 02:18 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,571
Mentioned: 456 Post(s)
Tagged: 0 Thread(s)
Quoted: 2330 Post(s)
Liked: 3002
Quote:
Originally Posted by Neo-XP View Post
However, I always test with D3D11 native, and I get different results by switching to D3D11 copy-back or other hardware decoders :

Code:
Fer15          : Atomic Blonde 6124 -> 28,58 10,88 11,79 16,22 10,88 10,88 11,79 16,22
D3D11 native   : Atomic Blonde 6124 -> 28,61 12,14 12,72 17,91 12,14 12,14 12,72 17,91
D3D11 copy-back: Atomic Blonde 6124 -> 28.59 10.91 11.84 16.28 10.91 10.90 11.84 16.28
DXVA2 native   : Atomic Blonde 6124 -> 27.76 10.93 11.87 16.38 10.93 10.93 11.87 16.38
DXVA2 copy-back: Atomic Blonde 6124 -> 28.59 10.91 11.84 16.29 10.91 10.91 11.84 16.29
None           : Atomic Blonde 6124 -> 28.59 10.91 11.84 16.29 10.91 10.91 11.84 16.29
After countless hours of debugging (ouch) I finally got this solved. I had to add another pixel shader pass to make "D3D11 native" get bit perfect results, compared to software decoding or copyback. So maybe "D3D11 native" might have become ever so slightly slower in the new build, but it shouldn't be much.

http://madshi.net/madVRhdrMeasure80.zip

This should be the same test build compared to the previous build, with only "D3D11 native" fixed to produce the exact same scene change measurements. No other changes yet.

Quote:
Originally Posted by Neo-XP View Post
But it should not matters, unless we merge results.
I'm not 100% sure about that, to be honest. It seems likely that you're right, but maybe we should double check, just to be safe? E.g. I've noticed that often in the previous build "D3D11 native" produced higher Metric2 values than "D3D11 copyback", but sometimes it was lower, so it *could* potentially have an influence on the number of overall misses.

Would you mind maybe just re-testing "tms + 0,1" and "tms + 1,0"? I guess if these two have the same number of misses after re-testing, we should be safe. Sorry for extra work...

Quote:
Originally Posted by Neo-XP View Post
Fer15 doc doesn't have much false positives though, I think that can be the reason why we don't get the same best algo.

When I begin testing, I always get fluctuating best algo until a certain amount of cases, then it stabilizes.
That makes sense. There's just this tiny doubt in the back of my head that maybe D3D11 native producing different Metric2 values could also play a role here? Probably not, but I'm not 100% sure.

Quote:
Originally Posted by ddgdl View Post
Madshi, is there any hope for black bar detection being available with D3D11 native?
What does this have to do with "Improving madVR HDR to SDR mapping for projector"? See title of this thread...

Quote:
Originally Posted by Manni01 View Post
- Please could you fix the dynamic clipping option [...]
- As Nevcairiel has done the necessary changes in LAV/jriver, please could you [...]
- can you share with us when you plan to work on the "looking ahead" [...]
As you know, I don't like to give out ETAs. I like to work on one thing at a time. So I usually get to things further down the to do list only after the things before that are done. I want/need to finish up scene detection first. After that I plan to work on / finish reaction speeds. After that I'll see what I consider most important then. Maybe looking head, maybe dynamic clipping, maybe something else, I don't know yet.
madshi is offline  
post #5904 of 6033 Old 05-17-2019, 08:43 PM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 606
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 298
Quote:
Originally Posted by madshi View Post
What does this have to do with "Improving madVR HDR to SDR mapping for projector"? See title of this thread...
Because...I forgot what thread I was in (highly ironic, given the number of times I have told people in this thread that this was only for beta testing)
stevenjw and Manni01 like this.
ddgdl is online now  
post #5905 of 6033 Old 05-18-2019, 12:46 AM
Advanced Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 835
Mentioned: 134 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 685
Quote:
Originally Posted by madshi View Post
Would you mind maybe just re-testing "tms + 0,1" and "tms + 1,0"? I guess if these two have the same number of misses after re-testing, we should be safe. Sorry for extra work...
The final result is still the same, but the good news is that there are even less misses with madVRhdrMeasure80:



Doc: https://www.mediafire.com/file/z8723..._V13.xlsx/file
stevenjw, madshi and Manni01 like this.
Neo-XP is offline  
post #5906 of 6033 Old 05-18-2019, 01:01 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,571
Mentioned: 456 Post(s)
Tagged: 0 Thread(s)
Quoted: 2330 Post(s)
Liked: 3002
Very nice - thanks for double checking...

Did I understand it correctly that you have *all* of Fer15's test scenes included in your sheet now? Or just some? Just wondering if we can consider your test results as the "law" now.
stevenjw and Neo-XP like this.
madshi is offline  
post #5907 of 6033 Old 05-18-2019, 01:25 AM
Advanced Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 835
Mentioned: 134 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 685
Quote:
Originally Posted by madshi View Post
Did I understand it correctly that you have *all* of Fer15's test scenes included in your sheet now? Or just some? Just wondering if we can consider your test results as the "law" now.
Only all the false positives, I'm missing a few real scenes.

But now that the values are pretty much similar between Fer15's and me, if I add the missing real scenes to my sheet, it goes like this:



Doc: https://www.mediafire.com/file/ua3b5..._V15.xlsx/file
madshi, Manni01 and Fer15 like this.
Neo-XP is offline  
post #5908 of 6033 Old 05-18-2019, 01:54 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,571
Mentioned: 456 Post(s)
Tagged: 0 Thread(s)
Quoted: 2330 Post(s)
Liked: 3002
Cool, thanks. So we can move on to the next stage now. (Probably) two more test builds to go, then we can retire this Excel sheet tests.
stevenjw, SOWK, SamuriHL and 6 others like this.
madshi is offline  
post #5909 of 6033 Old 05-18-2019, 12:00 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,571
Mentioned: 456 Post(s)
Tagged: 0 Thread(s)
Quoted: 2330 Post(s)
Liked: 3002
So here's the next test build:

http://madshi.net/madVRhdrMeasure81.zip

I didn't work on the fade in/out yet. This test build just modifies the "Metric2 algo" once more. This is probably going to be the last *quality* related Metric2 test build. Afterwards there's just going to be one more test build to maybe improve speed (if possible without losing too much detection quality).

You can now choose a minimum amount of pixels madVR looks at for Metric2. The default of the previous test build was 35%. You can choose a higher percentage. Doing so will help high APL scenes, but it will probably hurt other scenes. So this one could be tricky to choose a "good" value for. I've chosen those 35% to avoid some false positives in Fer15's test set. But maybe choosing a higher value could still be useful. Historically, IIRC Fer15 likes lower values here and Neo-XP higher values, so another reason why this could be tricky.

@Neo-XP , two questions:

1) Would it make sense to maybe add a separate "misses sum" line for high APL scenes and to exclude them from the other sums? Right now you're separating "false positive misses" vs "real scene misses". So there would be a 3rd line for "high APL misses". Doing that would allow us to check if switching from 35% to a higher value only helps high APL scenes, or if it could potentially help other scenes, as well. Generally, I'm not sure if we need to worry too much about high APL scenes, because Metric1 does a great job with those. On the other hand, it we can make Metric2 handle those better without any negative side effects, why not?

2) A good question is how to split the testing work between you and Fer15, since you now merged all his scenes into your sheet. He had some suggestions, e.g. one of you could test false positives, the other one real scenes. Or maybe he would simply continue to test "his" test set and you would skip those and then you could throw your numbers together? It's not important to me how you split the work, of course, you can decide for yourself. Just wouldn't want you both to duplicate each other's work, to not waste any of your time.
stevenjw, woofer, SOWK and 6 others like this.
madshi is offline  
post #5910 of 6033 Old 05-19-2019, 03:41 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 8,989
Mentioned: 307 Post(s)
Tagged: 0 Thread(s)
Quoted: 5345 Post(s)
Liked: 5382
@madshi

Do we need to remeasure our files following the D3D11 native fix in build 80, or is it only a fix in the live algo?

Thanks.

JVC Autocal Software V11 Calibration for 2019 Models
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is online now  
Sponsored Links
Advertisement
 
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

Tags
dynamic tone mapping , hdr , madvr , sdr , ton mapping

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