Improving Madvr HDR to SDR mapping for projector - Page 43 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 4754Likes
Reply
 
Thread Tools
post #1261 of 6032 Old 07-17-2018, 04:12 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
Yes, even comparing 30% vs 35%, the difference is quite small. 33.33% vs 35% should be very near. Of course I could do 35% + 60%, but I'd rather use 55.55%, because I think that's a more sane option.

If we can ever compromise on only one hue shifting config, I could see me upping the value to maybe 37.5% or something like that. But I don't think Javs and RioBar4U would be happy with that, would they?
Manni01 likes this.
madshi is offline  
Sponsored Links
Advertisement
 
post #1262 of 6032 Old 07-17-2018, 04:12 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: 5381
Quote:
Originally Posted by Fer15 View Post
I think the difference would be almost imperceptible tbh.
That’s exactly what I said

It’s psychological if anything... might help convert those who still prefer higher hue corrections and lead to only two options.
Fer15 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 offline  
post #1263 of 6032 Old 07-17-2018, 04:18 AM
Senior Member
 
Join Date: Jan 2018
Posts: 350
Mentioned: 67 Post(s)
Tagged: 0 Thread(s)
Quoted: 188 Post(s)
Liked: 219
Quote:
Originally Posted by Manni01 View Post
That’s exactly what I said

It’s psychological if anything... might help convert those who still prefer higher hue corrections and lead to only two options.

I know. I was concurring with you.
Manni01 likes this.
Fer15 is offline  
Sponsored Links
Advertisement
 
post #1264 of 6032 Old 07-17-2018, 04:37 AM
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 madshi View Post
Yes, even comparing 30% vs 35%, the difference is quite small. 33.33% vs 35% should be very near. Of course I could do 35% + 60%, but I'd rather use 55.55%, because I think that's a more sane option.

If we can ever compromise on only one hue shifting config, I could see me upping the value to maybe 37.5% or something like that. But I don't think Javs and RioBar4U would be happy with that, would they?
Yeah not at this time.

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 #1265 of 6032 Old 07-17-2018, 10:32 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
Hmmmm... It seems that choosing gamma 2.0 reduces saturation for you? Is that the main reason why you're using it?
I use it only with a target peak nits of 400, because the image is too dark 99% of the time at 400nits.
And I use 400nits, because the image seems overexposed at 200nits in a lot of situations (and even at 300nits sometimes) and too saturated.
Setting the gamma at 2.0 makes the image look brighter (without looking overexposed like when reducing the target peak nits) and less saturated (comparing again to the same "brightness" I get when changing the target peak nits too 200 or 300).

I am certain it is not the right approach, but the result is that the films looks good with this configuration and it is the closest I could get to the HDR tone mapped version of my LG TV, so...
Maybe the "arve custom curve" can help me to achieve the same result, but I do not know how to use it.

Also, I noticed that the "measure each frame's peak luminance" option helps a lot to get closer to the HDR of the LG TV:

300 nits gamma 2.2 / 400 nits gamma 2.0 + measure each frame's peak luminance


Interstellar - frame 7494 (00:05:13)

Quote:
Originally Posted by madshi View Post
I'm not sure if that's really as intended, though. I wonder if there's a bug in my code somewhere.
Anyway. When you say "gamma 2.0", I suppose you're talking about madVR's "this display is already calibrated to" setting?
In color & gamma -> enable gamma processing -> pure power curve 2.00, but I guess it is the same as the gamma setting in the calibration page.

Quote:
Originally Posted by madshi View Post
Do you also adjust your TV/monitor settings accordingly? Is your TV/monitor calibrated at all?
No, my TV and monitor are calibrated for gamma 2.2 and I did not change this.


Is it possible to detect if a video is HDR or not to make a profile in madVR?

Last edited by Neo-XP; 07-17-2018 at 01:03 PM.
Neo-XP is online now  
post #1266 of 6032 Old 07-17-2018, 01:25 PM
Member
 
RioBar4U's Avatar
 
Join Date: Dec 2011
Location: Switzerland
Posts: 60
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 12 Post(s)
Liked: 12
Quote:
Originally Posted by Javs View Post
Yeah not at this time.
Happier with 50%, but you know we Swiss people like a good compromise: <50% would be good enough....
RioBar4U is offline  
post #1267 of 6032 Old 07-17-2018, 04:09 PM
Advanced Member
 
hasta666's Avatar
 
Join Date: Apr 2016
Posts: 621
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 458 Post(s)
Liked: 151
Quote:
Originally Posted by Neo-XP View Post
Is it possible to detect if a video is HDR or not to make a profile in madVR?


Of course. Just look at the madVR thread at doom9’s.
hasta666 is online now  
post #1268 of 6032 Old 07-17-2018, 06:15 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
Quote:
Originally Posted by hasta666 View Post
Of course. Just look at the madVR thread at doom9’s.
It is down for a few days now, but I will try to get the information from a cached version.

EDIT: OK it is very easy:

Code:
if (HDR) "HDR" else "SDR"

Last edited by Neo-XP; 07-17-2018 at 06:25 PM.
Neo-XP is online now  
post #1269 of 6032 Old 07-17-2018, 06:39 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
Alright so now that most of us have commented on the preferences.

I am really excited to move on to things like active frame measuring and maybe some control on the BT2390 curves themselves.

Also something that may be able to measure black levels in the content actively, there are a small handful of films which have a higher black level than HDR level 64, it would be really excellent if MadVR could detect this on a 'very' long rolling average and adjust. Maybe very hard to do though effectively.
stevenjw, mavang, KoKlusz and 1 others 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 #1270 of 6032 Old 07-17-2018, 08:46 PM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
Quote:
Originally Posted by fhoech View Post
Humor me and reduce your display brightness so it produces 48 cd/m2 peak for SDR playback. Now you have an actual, valid comparison. If you find it too dim, you should calibrate to a higher peak for HDR playback.
I am not sure I follow. I have an LCD display, perhaps you were assuming I have a projector? Could you explain why would I want to reduce the peak screen brightness to 48 cd/m2 to have a valid comparison?

Quote:
Originally Posted by fhoech View Post
You can turn that off though (I never use it because it has some side effects, e.g. causes extreme desaturation in highly saturated blue hues).
You can't. I think you are referring to 'restore details in compressed highlights'. I am talking about reducing the highlight recovery in the first place. madVR is really too aggressive pulling highlights back trying to recover as much data there as possible, which reduces the contrast too much, and make things look too flat (including actors' face) compared to SDR BR. Maybe this is baked into BT.2390 I am not sure.

Quote:
Originally Posted by fhoech View Post
You can certainly use clipping for HDR to SDR conversion in madVR, but it won't produce a pleasing picture if target peak is too low (too much picture information is lost).
I've found clipping to be the best approximation so far in terms of approaching to SDR Bray in terms of colors (saturation and hue) as well as image contrast. Unfortunately there's quite a bit variance between titles. While The Last Jedi and Bladerunner looks great with clipping, Batman v Superman does not and shows too much clipping for highlights.
bcec is offline  
post #1271 of 6032 Old 07-18-2018, 01:50 AM
Member
 
((( atom )))'s Avatar
 
Join Date: Jan 2010
Posts: 75
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 28
I get the impression that some people have uncalibrated projectors/TVs and are unintentionally trying to compensate for that using the various parameters that are being worked on here. I really have no need to use any gamma other that 2.2 and I never saw faces look better in SDR that HDR.

No offence though, but that can be very misleading. When switching even from "Hardware calibrated to D65" to "HW cal. + 3DLUT" I see quite some differences in color and brightness distribution and it sometimes has similar effects to switching between the various parameters we have been testing. This suggests that these parameters might visually compensate for display errors when they are not meant to.

It really jumped out on me recently when I finally mapped 8 madVR profiles to buttons on my remote control for very seamless switching and included profiles without LUT correction.

I know that is very hard to address but we should take care of such things, since they is a lot of work put into this, especially on madshi's side.
((( atom ))) is offline  
post #1272 of 6032 Old 07-18-2018, 01:55 AM
Member
 
MDesigns's Avatar
 
Join Date: Dec 2016
Location: Finland
Posts: 153
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 121 Post(s)
Liked: 63
Quote:
Originally Posted by bcec View Post
I am not sure I follow. I have an LCD display, perhaps you were assuming I have a projector? Could you explain why would I want to reduce the peak screen brightness to 48 cd/m2 to have a valid comparison?

If you watch SDR with 120 cd/m2 and HDR with 120 cd/m2 peak brightness, there is no room for HDR highlights. You should have atleast double the peak brightness for HDR to get proper average brightness and leave some room for the hdr highlights. That is why you need so extreme settings, you cant(or shouldn't) watch sdr and hdr with the same peak brightness.

JVC DLA-RS500
MDesigns is online now  
post #1273 of 6032 Old 07-18-2018, 04:46 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: 5381
Quote:
Originally Posted by Javs View Post
Alright so now that most of us have commented on the preferences.

I am really excited to move on to things like active frame measuring and maybe some control on the BT2390 curves themselves.

Also something that may be able to measure black levels in the content actively, there are a small handful of films which have a higher black level than HDR level 64, it would be really excellent if MadVR could detect this on a 'very' long rolling average and adjust. Maybe very hard to do though effectively.
Steady, Madshi will move to the next stage when he’s ready! It’s not as if he has forgotten, we just have agreed to put this aside for now. I have A LOT of ideas regarding this, but I’m waiting for his green light to start discussing this again.

One thing that Madshi has confirmed was high on his list, as a first step, is a new variable that will allow us to select a different pixel shader profile according to the peak brightness of the content based on the metadata. It’s not just mastering max brightness, but a combination of various element similar to the algo I developped for the Vertex. So hopefully this will bring us closer to the functionality we have with the Vertex when using passthrough, to automatically adapt the conversion to the content’s brightness using profiles instead of changing profiles manually.

The rest will come as and when Madshi sees fit
stevenjw, neo_2009, Javs and 3 others 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; 07-18-2018 at 04:52 AM.
Manni01 is offline  
post #1274 of 6032 Old 07-18-2018, 04:57 AM
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
Steady, Madshi will move to the next stage when he’s ready! It’s not as if he has forgotten, we just have agreed to put this aside for now. I have A LOT of ideas regarding this, but I’m waiting for his green light to start discussing this again.

One thing that Madshi has confirmed was high on his list, as a first step, is a new variable that will allow us to select a different pixel shader profile according to the peak brightness of the content based on the metadata. It’s not just mastering max brightness, but a combination of various element similar to the algo I developped for the Vertex. So hopefully this will bring us closer to the functionality we have with the Vertex when using passthrough, to automatically adapt the conversion to the content’s brightness using profiles instead of changing profiles manually.

The rest will come as and when Madshi sees fit
Exciting times ahead indeed.
Manni01, arcspin and Neo-XP 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 #1275 of 6032 Old 07-18-2018, 06:27 AM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
Quote:
Originally Posted by MDesigns View Post
If you watch SDR with 120 cd/m2 and HDR with 120 cd/m2 peak brightness, there is no room for HDR highlights. You should have atleast double the peak brightness for HDR to get proper average brightness and leave some room for the hdr highlights. That is why you need so extreme settings, you cant(or shouldn't) watch sdr and hdr with the same peak brightness.
I don't follow. I am watching SDR with 120 cd/m2 on my LCD. madVR's tone mapping converts HDR to SDR, so I should still be able to watch the result of tone mapping at 120cd/m2 peak brightness.
bcec is offline  
post #1276 of 6032 Old 07-18-2018, 08:15 AM
Advanced Member
 
neo_2009's Avatar
 
Join Date: Mar 2009
Posts: 656
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 309 Post(s)
Liked: 225
Thumbs up

Quote:
Originally Posted by Manni01 View Post
The rest will come as and when Madshi sees fit
I just want to thank you all for the amazing work you are doing in prol of the community, as this allows regular users (without deep pockets), to have state-of-the-art processing that otherwise would only be possible spending lots of $$$. I will gladly contribute when @madshi will be available for donations.

Thank you all !!!
Onkyoman, Manni01, mkohman and 4 others like this.
neo_2009 is online now  
post #1277 of 6032 Old 07-18-2018, 10:58 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,278
Mentioned: 194 Post(s)
Tagged: 0 Thread(s)
Quoted: 939 Post(s)
Liked: 1372
Now that the three sub-modes wth percentage of hue-shift are selected, I am also looking forward to the dynamic mapping based on real time measurement of peak&histogramm of brightness.

It should bring a huge benefit for projector user. Especially for dark scenes which should be less dark in the future without compromising bright scenes :-)
Soulnight is offline  
post #1278 of 6032 Old 07-18-2018, 07:15 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
Ok I just realized what was going on with the saturation problem at 200nits.

It is the change from "madVRhdrTest10" to "madVRhdrTest11" with forced 100% saturation correction for Dumb that is causing this:

SDR Bluray / madVRhdrTest9 - Dumb / madVRhdrTest10 - Dumb - 100% hue / madVRhdrTest11 - Dumb - 100% hue 100% saturation / madVRhdrTest15 - Bern no surface



There was no banding problem with Dumb before introducing the 100% saturation correction:

Bern no surface / Dumb 100% hue



Edit:

And if you want that extra pop in colors, dumb 100% hue 50% saturation is already saturated enough at 200 nits:

Dumb 100% hue 50% saturation / Bern no surface


Javs likes this.

Last edited by Neo-XP; 07-18-2018 at 08:32 PM.
Neo-XP is online now  
post #1279 of 6032 Old 07-18-2018, 11:45 PM
Member
 
MDesigns's Avatar
 
Join Date: Dec 2016
Location: Finland
Posts: 153
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 121 Post(s)
Liked: 63
Quote:
Originally Posted by bcec View Post
I don't follow. I am watching SDR with 120 cd/m2 on my LCD. madVR's tone mapping converts HDR to SDR, so I should still be able to watch the result of tone mapping at 120cd/m2 peak brightness.
Yes you could do that, but then you are compressing ALL the HDR highlights to zero nits, because you already use the whole 0-120cd/m2 for SDR-range. I don't believe that is the intention of the feature, disregard of the name. It converts and tone maps the absolute HDR EOTF to relative SDR gamma, but it doesn't(have to) make the picture SDR.

JVC DLA-RS500
MDesigns is online now  
post #1280 of 6032 Old 07-19-2018, 10:55 AM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
I just started using the latest build, so I don't have a ton of impressions to share.

My question is about black and white clipping? Has anyone here had any luck using HDR10 test patterns?

I found a couple that seem to help, but I'd like to test again with a second display. The first pattern was a combined 8-bit/10-bit black clipping video that responded to both changes in target nits and gamma, although subtle in some cases. But it definitely seemed to help remove black crush on one display. 500 nits, for example, was crushing part of Bar 17 with a fixed gamma curve.

White clipping was more confusing. A 10-bit, 240-1000 nit clipping pattern produced different clipping depending on the target nits. A setting of 200 nits clipped early, while a setting of 1,000 nits didn't clip at all. I found a few values that clipped perfectly at 1,000 nits. Going to 4,000 nits, the ideal setting for 1,000 nits clipped at about 2,000 nits. A setting close to 4,000 nits didn't clip white at all. I see how desaturated colors become at higher nits, so the clipping might be expected.

I don't know what to make of this other than to assume 200 nits is overexposed for white (and not to mention flat), while 500 nits is crushing at least some of black. This would vary, of course, due to variability in calibrated gamma curves, a display's accuracy with a given gamma curve and the brightness of the display.

I have to say a big thanks for all of the effort madshi put into this I am particularly impressed that you would put the final say into the most dedicated users.

HDR -> SDR is a niche feature is to some, but I can't see myself upgrading to a 4K display for a while and I've managed to get an image thus far that looks close to what I would expect from a PQ HDR image. It is hard to complain about that. The three 1,000 nits titles I looked at or watched seem to pass the tone mapping test easily. Only the couple of 4,000 nits titles I've seen left me with any questions about grading or color accuracy. Can't really say without slowly watching a few more movies. I won't have a large sample size to draw from like others here, so I don't know if I have any objections or not.
Onkyoman is offline  
post #1281 of 6032 Old 07-19-2018, 07:12 PM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
Quote:
Originally Posted by MDesigns View Post
Yes you could do that, but then you are compressing ALL the HDR highlights to zero nits, because you already use the whole 0-120cd/m2 for SDR-range. I don't believe that is the intention of the feature, disregard of the name. It converts and tone maps the absolute HDR EOTF to relative SDR gamma, but it doesn't(have to) make the picture SDR.
That should be the point of HDR->SDR conversion. Essentially replicating tone mapping done by studios for Blurays. Assuming studios did their job correctly for UHD and BR, a perfect on the fly tone mapping algorithm should result in the exact look of the BR done by the studios, i.e. certain highlight information clipped/lost. Achieving such perfect on the fly tone mapping is unrealistic, however, cramming in all the available data into the narrow space will only result in loss of contrast and increased saturation, neither of which are natural.
bcec is offline  
post #1282 of 6032 Old 07-20-2018, 08:22 AM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
Quote:
Originally Posted by bcec View Post
That should be the point of HDR->SDR conversion. Essentially replicating tone mapping done by studios for Blurays. Assuming studios did their job correctly for UHD and BR, a perfect on the fly tone mapping algorithm should result in the exact look of the BR done by the studios, i.e. certain highlight information clipped/lost. Achieving such perfect on the fly tone mapping is unrealistic, however, cramming in all the available data into the narrow space will only result in loss of contrast and increased saturation, neither of which are natural.
You probably wouldn't end up with the SDR Blu-ray colors because the source was remastered by a colorist. They can't just turn up the brightness knob to make a 4K version, so that comparison is usually flawed.

Clipping is one approach to tone mapping, but it is not the best. The image can be brighter but a lot of detail will be lost and you still have to tone map the pixels that clip or they will likely shift hue.

If the discussion is now on tone mapping curves, the following video could be helpful in showing three different approaches:


Sony uses a curve that is similar to simple clipping. LG uses a curve similar to BT.2390. And Panasonic uses a curve that follows PQ for a bit before rolling off. Each produces a distinct look. All displays have changed its tone mapping since this video.

You can skip to 8:24 to get to the good part.
stevenjw, NoTechi and Neo-XP like this.
Onkyoman is offline  
post #1283 of 6032 Old 07-20-2018, 09:29 AM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
No response from anyone on black and white clipping?

I tried to adjust a second 8-bit, 1080p display and had a different experience. Despite being dimmer, it required a much higher target nits to not appear washed out. Differences in gamma curves between displays seem to make HDR to SDR difficult to calibrate. Differences in the way an LED with micro dimming and an emissive plasma handle black information could also be the issue.

White clipping to 1,000 nits was reasonably easy to calibrate, but the plasma clipped white far earlier after 1,000 nits than the LED. Black clipping was more difficult. I couldn't get the display to avoid clipping some of black, but it didn't look that bad during actual playback. I don't know if the black clipping pattern is actually useful or not. I did appear to get proper clipping with the LED.

Here are the patterns I used if anyone else is interested in trying this. I'd like to know if these patterns are useful or not when using an SDR gamma curve rather than true PQ:

HDR10 Test Patterns: Black and White Clipping

Lastly, I thought I'd share one image capture from my computer monitor/plasma:



The image is from Red Sparrow. Not one of my few rips, so you get the idea. The plasma was clipping black a bit when I took the image, so some contrast would be lost. But, to me, it is a pretty good conversion from DCI-P3 to BT.709. Great color.
Onkyoman is offline  
post #1284 of 6032 Old 07-21-2018, 11:05 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
Another comparison that shows also a problem (same problem?) with the saturation correction at 200 nits:

1000 nits




200 nits - Bern no surface / Dumb 100% hue 50% saturation



200 nits - Dumb 100% hue 0% saturation / Dumb 100% hue 0% saturation + measure each frame's peak luminance



You need to look at the clouds.

Source is "Samsung Wonderland Two HDR UHD 4K Demo" at > 2min.

Last edited by Neo-XP; 07-21-2018 at 01:07 PM.
Neo-XP is online now  
post #1285 of 6032 Old 07-21-2018, 11:08 AM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
@madshi

original 'Dumb Mode - Convert Gamut Late' was great.

BR:


Dumb Mode - Convert Gamut Late:

(settings)


New release without any hue shift:


50% global hue shift (still not enought):


75% global hue shift. Looks better but colors are still quite off and way too reddish, compared to slight greenish hue in BR, which 'Dumb Mode - Convert Gamut Late' almost successfully reproduced. Also, many scenes at this extreme hue shift show artifacts, which were not present with 'Dumb Mode - Convert Gamut Late'.


Can we get the old pure dumb mode again? That was so far has been the best tone mapping algo that reproduced the look that was closest to brays.

(Screenshots are from BvS at 02:37:39.157)
Attached Thumbnails
Click image for larger version

Name:	old-madvr.png
Views:	278
Size:	3.49 MB
ID:	2431096   Click image for larger version

Name:	bra.png
Views:	329
Size:	2.67 MB
ID:	2431098   Click image for larger version

Name:	new-madvr.png
Views:	281
Size:	5.12 MB
ID:	2431102   Click image for larger version

Name:	75-hue.png
Views:	277
Size:	3.39 MB
ID:	2431124   Click image for larger version

Name:	50-hue.png
Views:	272
Size:	3.37 MB
ID:	2431126  

Click image for larger version

Name:	without-settings.JPG
Views:	262
Size:	159.0 KB
ID:	2431158  

Last edited by bcec; 07-21-2018 at 11:29 AM.
bcec is offline  
post #1286 of 6032 Old 07-21-2018, 11:26 AM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
@madshi

Also, in the old release with 'Dumb Mode - Convert Gamut Late', it was restore details in compressed highlights that was causing the hue shift away from yellow tones:

Without 'Restore Details in Compressed Highlights'


Settings:


With 'Restore Details in Compressed Highlights'


Settings:
Attached Thumbnails
Click image for larger version

Name:	with-settings.JPG
Views:	271
Size:	164.3 KB
ID:	2431150   Click image for larger version

Name:	without-settings.JPG
Views:	268
Size:	159.0 KB
ID:	2431152   Click image for larger version

Name:	without restore.jpg
Views:	263
Size:	504.1 KB
ID:	2431154   Click image for larger version

Name:	with restore.jpg
Views:	270
Size:	509.5 KB
ID:	2431156  
bcec is offline  
post #1287 of 6032 Old 07-21-2018, 11:50 AM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
Quote:
Originally Posted by bcec View Post
@madshi

original 'Dumb Mode - Convert Gamut Late' was great.

BR:


Can we get the old pure dumb mode again? That was so far has been the best tone mapping algo that reproduced the look that was closest to brays.

(Screenshots are from BvS at 02:37:39.157)
It was already discussed in length. While that image looks attractive with a global hue shift, many examples have been found where the hue shift is unintended or random. There aren't many UHD Blu-ray reviews where the color of the UHD version has been said to match the SDR Blu-ray regardless of the review display, so I still don't understand how the SDR Blu-ray can used as a reference without making assumptions that might not be true.

Personally, I would only assume the brightness scenes would require the most correction in saturation or possibly hue. But given the bulk of the movie is in the lower nits region, enforcing a global hue correction instead of hue preservation would probably make the overall movie less accurate. Using catalogue titles as the reference might also be questionable if they were pumped up to sell UHD discs.
Onkyoman is offline  
post #1288 of 6032 Old 07-21-2018, 11:58 AM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
Quote:
Originally Posted by Onkyoman View Post
It was already discussed in length. While that image looks attractive with a global hue shift, many examples have been found where the hue shift is unintended or random. There aren't many UHD Blu-ray reviews where the color of the UHD version has been said to match the SDR Blu-ray regardless of the review display, so I still don't understand how the SDR Blu-ray can used as a reference without making assumptions that might not be true.

Personally, I would only assume the brightness scenes would require the most correction in saturation or possibly hue. But given the bulk of the movie is in the lower nits region, enforcing a global hue correction instead of hue preservation would probably make the overall movie less accurate. Using catalogue titles as the reference might also be questionable if they were pumped up to sell UHD discs.
When you quoted, you removed my key comment (and I am not sure if you actually read that or skimmed). Here it is for context:

Quote:
Originally Posted by bcec
75% global hue shift. Looks better but colors are still quite off and way too reddish, compared to slight greenish hue in BR, which 'Dumb Mode - Convert Gamut Late' almost successfully reproduced. Also, many scenes at this extreme hue shift show artifacts, which were not present with 'Dumb Mode - Convert Gamut Late'.
With regards to Bray's look, they are done by professional colorists. Although one can't expect 100% match, however I'd expect a general closeness. And my point is that the previous Dumb Mode - Convert Gamut Late has so far been the best approximator across all the titles I've tried. It also kept the hue perfectly in those bright colors, so long as 'restore details in compressed highlights' was not selected (which could be a madvr bug, I don't know), and did not need an arbitrary hue correction step we are tying to do now.

With regards to the video you provided before with respect to my comment about too much highlight recovery in madVR (and thanks for the link for the video), you can see clearly in the video that Sony's rendition is the best (in all the rest, Ben Affleck's face looks sun tan, and image looks too dark - even though it is supposed to be a crazy bright scene).

Losing details in highlights is not a bad thing. That's how we perceive brightness. Further, if you look at Brays, most highlights are always clipped, and there's a reason why professional colorists do that when they grade movies. Same thing with HDR photography. If you pull in too much highlights and increase Shadows, you end up with a completely unnatural looking image.

Last edited by bcec; 07-21-2018 at 12:08 PM.
bcec is offline  
post #1289 of 6032 Old 07-21-2018, 12:18 PM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
Sure, I could agree that different types of tone mapping curves should be available. I kind of like Panasonic's approach in that video the most, but it is said to create the most posterization. A customizable curve or different options for different types of curves would be nice.

Getting the brightness correct with an accurate gamma curve seems to be the most difficult part when going from a bright, absolute PQ curve to a relative SDR gamma curve. Things like skin tones look off if this is not adjusted.
Onkyoman is offline  
post #1290 of 6032 Old 07-21-2018, 12:23 PM
AVS Forum Special Member
 
Onkyoman's Avatar
 
Join Date: Aug 2007
Location: Canada
Posts: 1,579
Mentioned: 40 Post(s)
Tagged: 0 Thread(s)
Quoted: 893 Post(s)
Liked: 548
I gave up on restore details in compressed highlights because it appeared to be enhancing noise and grain. I'm not sure if you've noticed this.
Onkyoman is offline  
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