Improving Madvr HDR to SDR mapping for projector - Page 18 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 8381Likes
Reply
 
Thread Tools
post #511 of 8777 Old 04-15-2018, 05:45 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Quote:
Originally Posted by madshi View Post
Argh.


Ok, thanks.


Thanks - appreciated! It seems that the Lumagen doesn't desaturate much, if at all.

Could you try the green spear scene in Batman vs Superman? And the Mad Max scenes from Javs posts 496 and 499?

@Javs , could you please post the time codes for those 3?
All the test footage is here

https://www.dropbox.com/sh/o6louempl..._EV0tp3Ba?dl=0

Otherwise for Mad Max its 20:12

And Batman v Superman spear is 2:32:55
SamuriHL 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 online now  
Sponsored Links
Advertisement
 
post #512 of 8777 Old 04-15-2018, 05:56 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Quote:
Originally Posted by madshi View Post
What is the time code of the two Mad Max images you posted?

Hmmmm... I wonder if we should wait with posting more real world screenshots until we have found the reason for the banding in the other color bar test pattern you posted. I have zero banding in that test pattern on my PC, using HDR test build 5. So I wonder why you do have banding. Can you double check if dithering is enabled? Maybe try disabling 10bit? Maybe try software decoding? Maybe try disabling the 3dlut? Etc... Something must be wrong, not sure what.
Ok, I found it.

Sorry Madshi my bad with this one.

I was using MPCHC today since its really easy to screenshot and see whats happening.

Had a look at the colour bars in Kodi and yes it renders with no banding there, so please disregard the colour bar banding issue, thats my fault.

This is correct:


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 online now  
post #513 of 8777 Old 04-15-2018, 06:11 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Just to add to the above, that MadMax shot I posted earlier in post 496, I checked this with Kodi (which does display the above shot with no banding) and MPC-HC at the same time.

The Mad Max time code is 28:53 blink and you miss it.

But here is where there is no highlight banding for Ictcp - Dynamic mode.



Worst



Better but still there



And Xyy is a trainwreck

Worst



Dynamic, still very bad.



Oh, and dumb mode.

Fer15 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

Last edited by Javs; 04-15-2018 at 06:15 PM.
Javs is online now  
Sponsored Links
Advertisement
 
post #514 of 8777 Old 04-15-2018, 11:21 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 360
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 306 Post(s)
Liked: 285
Quote:
Originally Posted by madshi View Post
Thanks - appreciated! It seems that the Lumagen doesn't desaturate much, if at all.

Could you try the green spear scene in Batman vs Superman? And the Mad Max scenes from Javs posts 496 and 499?

@Javs , could you please post the time codes for those 3?
No, i think it doesn't desaturate at all. And that is what I always wondered about when I saw the screenshots in this thread.

I don't have Batman vs Superman at the moment. But I can try to find the Mad Max scenes later. When think about how the Intensity Mapping of the Radiance usually reproduces those kind of highlights, I guess the flames will have a white center with a more or less noticeable border to the outer colors. But let me try this later.

It would be very interesting to see, how your madVR test pattern would look like on the Radiance Pro. But since it isn't available as .ts or something similar, it's not possible for me to play it.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3
KarlKlammer is offline  
post #515 of 8777 Old 04-16-2018, 01:13 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 670
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 368 Post(s)
Liked: 353
Has anyone found real world content showing the blue-pink hue shift with dumb tone mapping? Since there is almost no scenario where something that is supposed to be blue will look "better" to the eye if it is pink (unlike the explosions in mad max, and maybe the spear in BvS), I wonder if that wouldn't be the better test image to use?
ddgdl is offline  
post #516 of 8777 Old 04-16-2018, 01:32 AM
Advanced Member
 
EvLee's Avatar
 
Join Date: Dec 2009
Posts: 709
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 406 Post(s)
Liked: 420
Quote:
Originally Posted by madshi View Post
I agree with you that the DaVinci Resolve image looks better in this case than madVR's tone mapping. I think there are 2 key differences:

1) The DaVinci Resolve image has lower saturation. But that's actually not madVR's fault. I'm very careful to not change the saturation at all (except for rare overbright pixels)! The guy doing the DaVinci encode must have manually lowered saturation. This is not something I should do automatically, IMHO, because it would make some movies look under-saturated then.

2) I think DaVince Resolve probably uses a different tone mapping curve which helps with the face details/shadows. This is an area we will be working on here in the upcoming weeks. But again, there's a good chance the guy doing the encoding manually adjusted things. Which is not what madVR users will want to do. They will want to use one setting for all movies. So I have to find a good tone mapping curve which looks decent with all kinds of different material. Of course a custom tailored curve for one specific movie has a good chance of beating a default curve which tries to look decent for all movies.
I'll offer some pointers from my experience.

Generally speaking, it is advisable not to preserve all the saturation when performing tone mapping. Tone mapping is about making tradeoffs in multiple dimensions of color appearance, and sacrificing some saturation allows you to better maintain the local relationship between colors as well as image detail in highlights. I have tested this quite a bit over the years with multiple colorists, DPs, etc... and after being presented the options we always settle on a tone mapping that sacrifices some saturation for lightness. The direction (increase vs decrease) of saturation correction depends on what tone mapping operation you are using. This has been supported by color science research (see: https://www.cl.cam.ac.uk/~rkm38/pdfs/mantiuk09cctm.pdf, and http://www.erikreinhard.com/papers/cic_2013.pdf).

Hue preserving tone mapping is pretty much always desirable. There are various ways to go about it. The thing to watch out for is what space you use to calculate hue, and how the color volume in that space transforms back into display RGB. Performing hue correction in a perceptual space like ICtCp and then converting back to RGB can cause colors to rotate into clipping, so you want to check in your display RGB space that nothing is getting clipped.
madshi and Manni01 like this.
EvLee is offline  
post #517 of 8777 Old 04-16-2018, 04:25 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
Quote:
Originally Posted by Javs View Post
All the test footage is here

https://www.dropbox.com/sh/o6louempl..._EV0tp3Ba?dl=0

Otherwise for Mad Max its 20:12

And Batman v Superman spear is 2:32:55
Thanks!

Quote:
Originally Posted by Javs View Post
I was using MPCHC today since its really easy to screenshot and see whats happening.

Had a look at the colour bars in Kodi and yes it renders with no banding there, so please disregard the colour bar banding issue, thats my fault.
Strange! I usually use MPC-HC for such tests, too, because as you say, it's easiest to do quick tests with it. I wonder what MPC-HC setting was causing the issue! Anyway, it's not important for me to know.

Quote:
Originally Posted by Javs View Post
Just to add to the above, that MadMax shot I posted earlier in post 496, I checked this with Kodi (which does display the above shot with no banding) and MPC-HC at the same time.

The Mad Max time code is 28:53 blink and you miss it.

But here is where there is no highlight banding for Ictcp - Dynamic mode.

And Xyy is a trainwreck

Oh, and dumb mode.
Dynamic ICtCp most looks best to my eyes here, I think, better than dumb mode, or what do you think?

Quote:
Originally Posted by KarlKlammer View Post
No, i think it doesn't desaturate at all. And that is what I always wondered about when I saw the screenshots in this thread.

I don't have Batman vs Superman at the moment. But I can try to find the Mad Max scenes later. When think about how the Intensity Mapping of the Radiance usually reproduces those kind of highlights, I guess the flames will have a white center with a more or less noticeable border to the outer colors. But let me try this later.

It would be very interesting to see, how your madVR test pattern would look like on the Radiance Pro. But since it isn't available as .ts or something similar, it's not possible for me to play it.
Yeah, I was hoping a user here would know how to encode my test pattern into a .ts or .mkv file, but so far nobody seems to know (or have the time to try). I suppose dumping my test pattern's raw data into a file and then somehow running it through x265 might work, but I don't know for sure if it's possible to explain to x265 which format my test pattern's raw data has.

Quote:
Originally Posted by ddgdl View Post
Has anyone found real world content showing the blue-pink hue shift with dumb tone mapping? Since there is almost no scenario where something that is supposed to be blue will look "better" to the eye if it is pink (unlike the explosions in mad max, and maybe the spear in BvS), I wonder if that wouldn't be the better test image to use?
Agreed, having a real world sample showing pink instead of blue would be great. Haven't found such a sample yet, though. Such samples might already exist, but users might not notice it too much (myself included) while watching the movie, without specifically looking out for it.

Quote:
Originally Posted by EvLee View Post
I'll offer some pointers from my experience.

Generally speaking, it is advisable not to preserve all the saturation when performing tone mapping. Tone mapping is about making tradeoffs in multiple dimensions of color appearance, and sacrificing some saturation allows you to better maintain the local relationship between colors as well as image detail in highlights. I have tested this quite a bit over the years with multiple colorists, DPs, etc... and after being presented the options we always settle on a tone mapping that sacrifices some saturation for lightness. The direction (increase vs decrease) of saturation correction depends on what tone mapping operation you are using. This has been supported by color science research (see: https://www.cl.cam.ac.uk/~rkm38/pdfs/mantiuk09cctm.pdf, and http://www.erikreinhard.com/papers/cic_2013.pdf).

Hue preserving tone mapping is pretty much always desirable. There are various ways to go about it. The thing to watch out for is what space you use to calculate hue, and how the color volume in that space transforms back into display RGB. Performing hue correction in a perceptual space like ICtCp and then converting back to RGB can cause colors to rotate into clipping, so you want to check in your display RGB space that nothing is getting clipped.
Thank you very much, I appreciate your feedback!! That 2nd paper you linked to looks promising, because its processing is very near to what madVR is already doing, anyway, so I should easily be able to implement the algorithm suggested there.
Manni01 likes this.
madshi is offline  
post #518 of 8777 Old 04-16-2018, 04:54 AM
AVS Forum Special Member
 
markmon1's Avatar
 
Join Date: Dec 2006
Posts: 7,008
Mentioned: 130 Post(s)
Tagged: 0 Thread(s)
Quoted: 5844 Post(s)
Liked: 3919
Quote:
Originally Posted by madshi View Post
Yeah, I was hoping a user here would know how to encode my test pattern into a .ts or .mkv file, but so far nobody seems to know (or have the time to try). I suppose dumping my test pattern's raw data into a file and then somehow running it through x265 might work, but I don't know for sure if it's possible to explain to x265 which format my test pattern's raw data has.
Where is your test pattern and what format is it in?

Video: JVC RS4500 135" screen in pure black room no light, htpc nvidia 1080ti.
Audio: Anthem mrx720 running 7.1.4, McIntosh MC-303, MC-152, B&W 802d3 LR, B&W HTM1D3 center, B&W 805d3 surround, B&W 702S2 rear, B&W 706s2 x 4 shelf mounted for atmos, Infinite Baffle Subs 4x15 fi audio running on behringer ep4000 + 2x12 fi audio mounted in main chair firing into back.
markmon1 is offline  
post #519 of 8777 Old 04-16-2018, 05:03 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
It's a 32bit DirectShow filter which calculates pixels via software and then outputs them as 16bit 4:4:4 YCbCr pixels. If it helps, I can provide the raw pixel data?
madshi is offline  
post #520 of 8777 Old 04-16-2018, 05:14 AM
AVS Forum Special Member
 
markmon1's Avatar
 
Join Date: Dec 2006
Posts: 7,008
Mentioned: 130 Post(s)
Tagged: 0 Thread(s)
Quoted: 5844 Post(s)
Liked: 3919
Quote:
Originally Posted by madshi View Post
It's a 32bit DirectShow filter which calculates pixels via software and then outputs them as 16bit 4:4:4 YCbCr pixels. If it helps, I can provide the raw pixel data?
If that's what it does, can it be loaded into an avs script with directshowsource?

Video: JVC RS4500 135" screen in pure black room no light, htpc nvidia 1080ti.
Audio: Anthem mrx720 running 7.1.4, McIntosh MC-303, MC-152, B&W 802d3 LR, B&W HTM1D3 center, B&W 805d3 surround, B&W 702S2 rear, B&W 706s2 x 4 shelf mounted for atmos, Infinite Baffle Subs 4x15 fi audio running on behringer ep4000 + 2x12 fi audio mounted in main chair firing into back.
markmon1 is offline  
post #521 of 8777 Old 04-16-2018, 05:23 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
I've just tried but it doesn't seem to work, for unknown reasons.
madshi is offline  
post #522 of 8777 Old 04-16-2018, 05:37 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 670
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 368 Post(s)
Liked: 353
I wonder if some algorithmic search through a film like BvS or Mad Max could be done, looking for a pink pixel next to blue, which would at least give a start point for searching for such a blue to pink hue shift. May be more trouble than its worth
ddgdl is offline  
post #523 of 8777 Old 04-16-2018, 05:47 AM
AVS Forum Special Member
 
markmon1's Avatar
 
Join Date: Dec 2006
Posts: 7,008
Mentioned: 130 Post(s)
Tagged: 0 Thread(s)
Quoted: 5844 Post(s)
Liked: 3919
Quote:
Originally Posted by madshi View Post
I've just tried but it doesn't seem to work, for unknown reasons.
You can output this to a raw file? Like a yuv file or such? I can see if any of the ffmpeg builds I have support it. It looks like maybe pix_fmt yuv444p16le may read it.

If this is helpful you can try ffmpeg: https://www.convertthisfile.com/files/ffmpeg.exe
Something like ffmpeg -f rawvideo -pix_fmt yuv444p16le -s:v 3840x2160 -r 25 -i inputfile.yuv -c:v libx265 outputfile.mkv
where s:v is resolution and -r 25 is frame rate

ffmpeg -pix_fmts will list all the formats.

Video: JVC RS4500 135" screen in pure black room no light, htpc nvidia 1080ti.
Audio: Anthem mrx720 running 7.1.4, McIntosh MC-303, MC-152, B&W 802d3 LR, B&W HTM1D3 center, B&W 805d3 surround, B&W 702S2 rear, B&W 706s2 x 4 shelf mounted for atmos, Infinite Baffle Subs 4x15 fi audio running on behringer ep4000 + 2x12 fi audio mounted in main chair firing into back.

Last edited by markmon1; 04-16-2018 at 05:55 AM.
markmon1 is offline  
post #524 of 8777 Old 04-16-2018, 02:26 PM
Senior Member
 
Join Date: Jan 2018
Posts: 410
Mentioned: 109 Post(s)
Tagged: 0 Thread(s)
Quoted: 226 Post(s)
Liked: 382
I’ve been following this thread out of interest and would just like to take a moment to express my appreciation for all the work that you guys are doing.

I’m not sure if you’ll find the following useful to the current discussion or not, but I would just like to provide a few more examples of the differences between the scientific and dumb mode when playing back further examples of real content.


I noticed that there were some UHD titles, which appeared to be yielding some questionable tone mapping effects when the scientific approach was enabled. Some of the Nolan UHD collection struggled in particular (The Prestige, Batman Begins, Inception). As you can see from the screens, Caucasian skin tones took on a redder/ more golden effect when the scientific approach was used. Furthermore, whites - such as the clouds in the sky, or snow also became a warm white to the point that they appeared almost a light yellow.

Some of the screens are frame accurate. Others are a couple of frames out*, but there is no shift in the colour ensuing from not being frame specific from that particular scene (* I don’t know if it’s a bug or not, but if I try to go to a specific frame when using an m2ts file in MadVR, most of the time, the frame that is rendered is a blocky mess. I would have to demux the file and switch it to an .mkv container in order to avoid that problem).


1080p – scientific approach – dumb approach

The Prestige:




















Batman Begins:






Inception:






Other titles also suffered with white flesh tones, with the result that skin tones had a redder/ more golden finish:

The Mummy (1999):








Starship Troopers:








La La Land:




Javs and mmdd like this.
Fer15 is offline  
post #525 of 8777 Old 04-16-2018, 02:26 PM
Senior Member
 
Join Date: Jan 2018
Posts: 410
Mentioned: 109 Post(s)
Tagged: 0 Thread(s)
Quoted: 226 Post(s)
Liked: 382
Murder on the Orient Express (2017):




Ghost Busters (1984):




Apollo 13:




Blade Runner:
Scientific v dumb



Wonder Woman:








Thor: Ragnorak








Star Trek: Into Darkness




Hell or High Water


Javs and mmdd like this.
Fer15 is offline  
post #526 of 8777 Old 04-16-2018, 02:26 PM
Senior Member
 
Join Date: Jan 2018
Posts: 410
Mentioned: 109 Post(s)
Tagged: 0 Thread(s)
Quoted: 226 Post(s)
Liked: 382
One example where it is difficult to tell between the scientific/dumb settings is Coco in terms of which is better:







The screens that were taken with the scientific approach enabled, were from either the latest official release build, or with the scientific settings enabled in the latest test build. Those taken with the dumb settings were done with the latest test build. I also included the 1080p SDR Blu-ray for a few of them.

Just my two cents, even if the dumber approach is not scientifically accurate or yields erroneous results in test patterns, it appears that Hollywood has catered for the transfers of UHD content with more amenable results when the dumb version is applied.
madshi, Javs and bcec like this.

Last edited by Fer15; 04-16-2018 at 04:04 PM.
Fer15 is offline  
post #527 of 8777 Old 04-16-2018, 02:34 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 360
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 306 Post(s)
Liked: 285
I think Javs has a talent to find scenes that put every tone mapping to shame.

OK, here are screenshots of the spear and two from Mad Max. I took two pictures from each scene. One is with the Radiance Pro Intensity Mapping (IM) and one with a LightSpace ST.2084-LUT (LS).
IM makes it almost impossible to capture the spear right. But the difference between yellow and green spots is really too subtle to catch.
The first scene from Mad Max doesn't look that horrible with IM in reality. But the odd color reproduction of the lightning and flame is clearly visible.
And the pink smoke in the second Mad Max picture might be a little bit off too.
Attached Thumbnails
Click image for larger version

Name:	Spear_IM.JPG
Views:	72
Size:	57.3 KB
ID:	2391736   Click image for larger version

Name:	Spear_LS.JPG
Views:	73
Size:	58.6 KB
ID:	2391738   Click image for larger version

Name:	MadMax1_IM.JPG
Views:	78
Size:	96.4 KB
ID:	2391742   Click image for larger version

Name:	MadMax1_LS.JPG
Views:	70
Size:	91.9 KB
ID:	2391748   Click image for larger version

Name:	MadMax2_IM.JPG
Views:	67
Size:	126.8 KB
ID:	2391750  

Click image for larger version

Name:	MadMax2_LS.JPG
Views:	65
Size:	135.5 KB
ID:	2391752  
Soulnight and Javs like this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3
KarlKlammer is offline  
post #528 of 8777 Old 04-16-2018, 03:13 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
Quote:
Originally Posted by EvLee View Post
Generally speaking, it is advisable not to preserve all the saturation when performing tone mapping. Tone mapping is about making tradeoffs in multiple dimensions of color appearance, and sacrificing some saturation allows you to better maintain the local relationship between colors as well as image detail in highlights. I have tested this quite a bit over the years with multiple colorists, DPs, etc... and after being presented the options we always settle on a tone mapping that sacrifices some saturation for lightness. The direction (increase vs decrease) of saturation correction depends on what tone mapping operation you are using. This has been supported by color science research (see: https://www.cl.cam.ac.uk/~rkm38/pdfs/mantiuk09cctm.pdf, and http://www.erikreinhard.com/papers/cic_2013.pdf).
I've tried the algorithm suggested in the 2nd paper - but using it doesn't produce the slighest visible change for me! I'm wondering if this is due to either:

a) Maybe it's a problem that I use ICtCp instead of IPT? But in theory ICtCp should be superior to IPT.

b) Maybe my tone mapping is already handling saturation in the way the algorithm aims for? FWIW, I'm doing tone mapping in ICtCp, and I am already doing a desaturation according to the way Dolby and SMPTE 2390 suggest. So maybe there's simply nothing left to do for the algorithm? I'm totally not sure about this, though.

Quote:
Originally Posted by EvLee View Post
Hue preserving tone mapping is pretty much always desirable. There are various ways to go about it. The thing to watch out for is what space you use to calculate hue, and how the color volume in that space transforms back into display RGB. Performing hue correction in a perceptual space like ICtCp and then converting back to RGB can cause colors to rotate into clipping, so you want to check in your display RGB space that nothing is getting clipped.
Yes, in highest quality mode I'm doing everything in ICtCp, and it's a bitch to work with. I think I have the clipping problems solved, but it's costing quite a bit of GPU performance.

Quote:
Originally Posted by markmon1 View Post
You can output this to a raw file? Like a yuv file or such? I can see if any of the ffmpeg builds I have support it. It looks like maybe pix_fmt yuv444p16le may read it.

If this is helpful you can try ffmpeg: https://www.convertthisfile.com/files/ffmpeg.exe
Something like ffmpeg -f rawvideo -pix_fmt yuv444p16le -s:v 3840x2160 -r 25 -i inputfile.yuv -c:v libx265 outputfile.mkv
where s:v is resolution and -r 25 is frame rate

ffmpeg -pix_fmts will list all the formats.
Thanks, I'll give that a try, when I find some time.

Quote:
Originally Posted by Fer15 View Post
I’ve been following this thread out of interest and would just like to take a moment to express my appreciation for all the work that you guys are doing.

I’m not sure if you’ll find the following useful to the current discussion or not, but I would just like to provide a few more examples of the differences between the scientific and dumb mode when playing back further examples of real content.

I noticed that there were some UHD titles, which appeared to be yielding some questionable tone mapping effects when the scientific approach was enabled. Some of the Nolan UHD collection struggled in particular (The Prestige, Batman Begins, Inception). As you can see from the screens, Caucasian skin tones took on a redder/ more golden effect when the scientific approach was used. Furthermore, whites - such as the clouds in the sky, or snow also became a warm white to the point that they appeared almost a light yellow.
Thanks for the comparison screenshots! Could you please provide timecodes for each of these screenshots? Without timecodes it would be *really* painful for me to analyze all these frames, because I'd have no easy way to double check this myself. I suppose these are all USA releases? Usually they have the same color grading as releases from other countries, but sometimes the color grading might differ, so it'd be good to know.

Quote:
Originally Posted by KarlKlammer View Post
I think Javs has a talent to find scenes that put every tone mapping to shame.

OK, here are screenshots of the spear and two from Mad Max. I took two pictures from each scene. One is with the Radiance Pro Intensity Mapping (IM) and one with a LightSpace ST.2084-LUT (LS).
IM makes it almost impossible to capture the spear right. But the difference between yellow and green spots is really too subtle to catch.
The first scene from Mad Max doesn't look that horrible with IM in reality. But the odd color reproduction of the lightning and flame is clearly visible.
And the pink smoke in the second Mad Max picture might be a little bit off too.
Thank you for the screenshots! Ok, so we can conclude that:

1) Lumagen's IM does not desaturate at all. It's the same approach which madVR uses if you select the option "100% luminance reduction, 0% saturation reduction". IMHO that's not a good choice, to be honest. As you can see, it produces highlight detail loss in some scenes. Ok, these scenes are extreme cases, but they do occur.

2) LightSpace seems to use "dumb/naive" color handling, which is surprising to me.
madshi is offline  
post #529 of 8777 Old 04-16-2018, 03:56 PM
Senior Member
 
Join Date: Jan 2018
Posts: 410
Mentioned: 109 Post(s)
Tagged: 0 Thread(s)
Quoted: 226 Post(s)
Liked: 382
Quote:
Originally Posted by madshi View Post


Thanks for the comparison screenshots! Could you please provide timecodes for each of these screenshots? Without timecodes it would be *really* painful for me to analyze all these frames, because I'd have no easy way to double check this myself. I suppose these are all USA releases? Usually they have the same color grading as releases from other countries, but sometimes the color grading might differ, so it'd be good to know.


.
No problem. Here you go.

Some are European releases, some are American sources. Most of the UHD titles have the same transfer, there are only a few where the grading is different i.e. Blade Runner 2049, and La La Land (I think the Japanese UHD release had different colour grading to Europe/USA, but I didn't use that source for the screens)



The Prestige:

00:10:50
01:21:40
00:51:12

Batman Begins:
00:49:44

Inception:
00:41:22

The Mummy (1999):
00:08:21
00:15:05


Starship Troopers:
00:03:12
00:14:34

La La Land
00:29:08

Murder on the Orient Express (2017):
00:21:54

Ghostbusters (1984):
00:33:51

Apollo 13:
00:35:28

Blade Runner:
00:17:21

Wonder Woman:
01:58:05
00:21:02

Thor: Ragnorak
00:02:52
00:04:33

Star Trek: Into Darkness
00:05:36

Hell or High Water
00:41:47


Also, I didn't include this one at 00:40:51 (it's got a nice neon sign)
dumb vs scientific













Coco:
00:14:20


One other image that might be of interest: Harry Potter and the Goblet of Fire at 02:08:20. A big difference between scientific vs dumb












Last edited by Fer15; 04-17-2018 at 11:00 AM.
Fer15 is offline  
post #530 of 8777 Old 04-16-2018, 04:12 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 360
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 306 Post(s)
Liked: 285
Quote:
Originally Posted by madshi View Post
Thank you for the screenshots! Ok, so we can conclude that:

1) Lumagen's IM does not desaturate at all. It's the same approach which madVR uses if you select the option "100% luminance reduction, 0% saturation reduction". IMHO that's not a good choice, to be honest. As you can see, it produces highlight detail loss in some scenes. Ok, these scenes are extreme cases, but they do occur.
Highlight detail loss does occur. And there are always a few scenes in a movie where that can be noticed.

Quote:
2) LightSpace seems to use "dumb/naive" color handling, which is surprising to me.
I was under the impression, that you need realtime processing to do smarter color handling. LightSpace tone mapping is based on static 1D/3D-LUTs.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3
KarlKlammer is offline  
post #531 of 8777 Old 04-16-2018, 04:42 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Quote:
Originally Posted by KarlKlammer View Post
I think Javs has a talent to find scenes that put every tone mapping to shame.

OK, here are screenshots of the spear and two from Mad Max. I took two pictures from each scene. One is with the Radiance Pro Intensity Mapping (IM) and one with a LightSpace ST.2084-LUT (LS).
IM makes it almost impossible to capture the spear right. But the difference between yellow and green spots is really too subtle to catch.
The first scene from Mad Max doesn't look that horrible with IM in reality. But the odd color reproduction of the lightning and flame is clearly visible.
And the pink smoke in the second Mad Max picture might be a little bit off too.
Wow, sorry to say but that kinda closes the book on the Lumagen Intensity Mapping in my view. I wouldnt be using that.

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 online now  
post #532 of 8777 Old 04-16-2018, 04:55 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Looking at all those screenshots... Seems like the scientific mode is arguably adding a bit too much saturation, or, its bringing it out a little too much, it seems a little overcooked to me to be honest. Is there a way to bring that under control a little? The faces are looking generally unnatural in most of those.

It also looks like the colour white is somehow being changed in colour temp.

Im wondering if there is a way to verify that these modes are not shifting colour saturation in any way past what is meant to be there, maybe we should be measuring these to see if they are adhering to the luminance and saturation specs for a correct calibration?

I suppose I could measure the Masciola BT2020 single colour slides since thats what its designed to do, but scientific algo doesnt display it properly they would probably be desaturated.

Any ideas?

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 online now  
post #533 of 8777 Old 04-17-2018, 02:16 PM
Advanced Member
 
EvLee's Avatar
 
Join Date: Dec 2009
Posts: 709
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 406 Post(s)
Liked: 420
Quick explanation of the golden/red comments seen with skin tones. Natural skin tones, before being tone mapped, will exhibit some variation in saturation but more so in lightness due to the way light falls across its contours. With bright skin tones, if you preserve saturation and focus solely on compressing lightness, the effect is the relative color separation in skin tones is reduced resulting in a flattened appearance. Additionally, we judge the color of surfaces relative to inferred illumination. Basically, the color of a material is its reflectance times the illumination, and our visual system is constantly trying to adapt to illumination to determine the base color of the material. When the highlights get compressed in lightness, we judge the illumination to be darker and this in turns increases the perceived chroma of the surface. In the case of skin tones, this makes them look redder. Rather than compressing only lightness and preserving all of saturation, it's better to balance lightness compression with a degree of desaturation. Doing so will preserve more of the detail and contours, as well as reduce the tendency of caucasian faces to look sun-burnt after tone mapping.
Manni01 likes this.

Last edited by EvLee; 04-17-2018 at 02:19 PM.
EvLee is offline  
post #534 of 8777 Old 04-17-2018, 04:11 PM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
Quote:
Originally Posted by Fer15 View Post
No problem. Here you go.
Thanks!

Quote:
Originally Posted by Fer15 View Post
One other image that might be of interest: Harry Potter and the Goblet of Fire at 02:08:20. A big difference between scientific vs dumb
Hmmmm... In which order are those? You've posted 3 Goblet of Fire comparison screenshots. In the first 2 of those the first image is more green-ish and the 2nd image is white-ish. However, in the third comparison, it's the first image which is more white-ish and the 2nd image is more green-ish. So which of these is scientific and which is dumb?

Quote:
Originally Posted by KarlKlammer View Post
I was under the impression, that you need realtime processing to do smarter color handling. LightSpace tone mapping is based on static 1D/3D-LUTs.
You can't do proper color handling when trying to do it in separate R/G/B 1DLUTs. However, you certainly *can* do smarter color handling inside of a big 3DLUT.

Quote:
Originally Posted by Javs View Post
Looking at all those screenshots... Seems like the scientific mode is arguably adding a bit too much saturation, or, its bringing it out a little too much, it seems a little overcooked to me to be honest. Is there a way to bring that under control a little? The faces are looking generally unnatural in most of those.

It also looks like the colour white is somehow being changed in colour temp.

Im wondering if there is a way to verify that these modes are not shifting colour saturation in any way past what is meant to be there, maybe we should be measuring these to see if they are adhering to the luminance and saturation specs for a correct calibration?

I suppose I could measure the Masciola BT2020 single colour slides since thats what its designed to do, but scientific algo doesnt display it properly they would probably be desaturated.
Properly analyzing this is really difficult. In theory, madVR should not touch pixels below diffuse white much (BT.2390) or at all (arve). So I'm not sure how madVR could be reponsible for incorrectly boosting or changing saturation for such pixels. But then, the whole HDR -> SDR conversion also involves converting PQ to Gamma, maybe (depending on settings) gamut mapping BT.2020 to BT.709 etc. Maybe something goes wrong there. But I have my doubts. 1080p Blu-Rays can't really be taken as a reference to how a tone mapped HDR Blu-Ray should look like. Some UHD HDR Blu-Ray are mastered with much higher saturation levels than the 1080p Blu-Rays, some are not. So it's really difficult to judge.

Quote:
Originally Posted by EvLee View Post
Quick explanation of the golden/red comments seen with skin tones. Natural skin tones, before being tone mapped, will exhibit some variation in saturation but more so in lightness due to the way light falls across its contours. With bright skin tones, if you preserve saturation and focus solely on compressing lightness, the effect is the relative color separation in skin tones is reduced resulting in a flattened appearance. Additionally, we judge the color of surfaces relative to inferred illumination. Basically, the color of a material is its reflectance times the illumination, and our visual system is constantly trying to adapt to illumination to determine the base color of the material. When the highlights get compressed in lightness, we judge the illumination to be darker and this in turns increases the perceived chroma of the surface. In the case of skin tones, this makes them look redder. Rather than compressing only lightness and preserving all of saturation, it's better to balance lightness compression with a degree of desaturation. Doing so will preserve more of the detail and contours, as well as reduce the tendency of caucasian faces to look sun-burnt after tone mapping.
Don't skin tone pixels usually have low(ish) nits levels? The custom arve curves don't compress pixels below a certain nits range *at all*. madVR's default BT.2390 tone mapping does compress pixels quite a bit earlier, but only very slightly. So I'm wondering if such strong saturation differences as seen in recent screenshots can be explained this way? I mean we're not talking about 1000nits+ highlight pixels here, or are we? Haven't really had a chance to analyze Fer15's screenshots in detail yet, though.
madshi is offline  
post #535 of 8777 Old 04-17-2018, 04:47 PM
Senior Member
 
Join Date: Jan 2018
Posts: 410
Mentioned: 109 Post(s)
Tagged: 0 Thread(s)
Quoted: 226 Post(s)
Liked: 382
Quote:
Originally Posted by madshi View Post
Thanks!


Hmmmm... In which order are those? You've posted 3 Goblet of Fire comparison screenshots. In the first 2 of those the first image is more green-ish and the 2nd image is white-ish. However, in the third comparison, it's the first image which is more white-ish and the 2nd image is more green-ish. So which of these is scientific and which is dumb?

Sorry about that. The first two are in the correct order. I got the third set mixed up.

All of the screenshots with the greener colour have the scientific settings enabled. Those with the whiter colour have the dumb settings enabled.
Fer15 is offline  
post #536 of 8777 Old 04-17-2018, 06:47 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Quote:
Originally Posted by madshi View Post
Properly analyzing this is really difficult. In theory, madVR should not touch pixels below diffuse white much (BT.2390) or at all (arve). So I'm not sure how madVR could be reponsible for incorrectly boosting or changing saturation for such pixels. But then, the whole HDR -> SDR conversion also involves converting PQ to Gamma, maybe (depending on settings) gamut mapping BT.2020 to BT.709 etc. Maybe something goes wrong there. But I have my doubts. 1080p Blu-Rays can't really be taken as a reference to how a tone mapped HDR Blu-Ray should look like. Some UHD HDR Blu-Ray are mastered with much higher saturation levels than the 1080p Blu-Rays, some are not. So it's really difficult to judge.
Yeah and since the Arve and BT2390 options both show this saturation difference/disparity when cycling between scientific and niave modes its kind of odd I think that the effects are very visible well below 100 nits content.

Skin tones and faces generally especially the ones which have been posted will be well under 100 nits in brightness, probably about 50 actually. That has also been the case in my HDR Waveform analysis thread, so there has gotta be something going on because this is a rather large shift in saturation it seems. At first with HDR demo material I did think this looks cool, but with graded film content its clear that this may potentially be moving into the category of running a bit hot somehow.

Thats why I am just wondering if there is a way we can measure this? Surely if we measure tone mapped colour primaries we can find out if the colour is correct and if the saturation is right.

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 online now  
post #537 of 8777 Old 04-17-2018, 07:00 PM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,968
Mentioned: 216 Post(s)
Tagged: 0 Thread(s)
Quoted: 4161 Post(s)
Liked: 7417
Quote:
Originally Posted by KarlKlammer View Post
I think Javs has a talent to find scenes that put every tone mapping to shame.

OK, here are screenshots of the spear and two from Mad Max. I took two pictures from each scene. One is with the Radiance Pro Intensity Mapping (IM) and one with a LightSpace ST.2084-LUT (LS).
IM makes it almost impossible to capture the spear right. But the difference between yellow and green spots is really too subtle to catch.
The first scene from Mad Max doesn't look that horrible with IM in reality. But the odd color reproduction of the lightning and flame is clearly visible.
And the pink smoke in the second Mad Max picture might be a little bit off too.
May I have the exact time stamps for these scenes? I want to check them with my Lumagen just to make sure this isn't a settings issue as I get wonky results if the settings aren't right, so I want to be sure that isn't what is causing these problems.

My Home Theater UPDATED NOV 2019
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is offline  
post #538 of 8777 Old 04-17-2018, 08:23 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,064
Mentioned: 501 Post(s)
Tagged: 0 Thread(s)
Quoted: 6937 Post(s)
Liked: 6661
Quote:
Originally Posted by Kris Deering View Post
May I have the exact time stamps for these scenes? I want to check them with my Lumagen just to make sure this isn't a settings issue as I get wonky results if the settings aren't right, so I want to be sure that isn't what is causing these problems.
The Mad Max time code is 28:53 blink and you miss it.

And Batman v Superman spear is 2:32:55
Kris Deering 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 online now  
post #539 of 8777 Old 04-17-2018, 11:28 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 360
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 306 Post(s)
Liked: 285
Quote:
Originally Posted by Kris Deering View Post
May I have the exact time stamps for these scenes? I want to check them with my Lumagen just to make sure this isn't a settings issue as I get wonky results if the settings aren't right, so I want to be sure that isn't what is causing these problems.
Additionally there is 20:12 for Mad Max.

Seriously, I would be happy if it were a settings issue.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3
KarlKlammer is offline  
post #540 of 8777 Old 04-18-2018, 01:56 AM
AVS Forum Special Member
 
madshi's Avatar
 
Join Date: May 2005
Posts: 7,899
Mentioned: 612 Post(s)
Tagged: 0 Thread(s)
Quoted: 2822 Post(s)
Liked: 4053
Quote:
Originally Posted by Javs View Post
Yeah and since the Arve and BT2390 options both show this saturation difference/disparity when cycling between scientific and niave modes its kind of odd I think that the effects are very visible well below 100 nits content.

Skin tones and faces generally especially the ones which have been posted will be well under 100 nits in brightness, probably about 50 actually. That has also been the case in my HDR Waveform analysis thread, so there has gotta be something going on because this is a rather large shift in saturation it seems. At first with HDR demo material I did think this looks cool, but with graded film content its clear that this may potentially be moving into the category of running a bit hot somehow.

Thats why I am just wondering if there is a way we can measure this? Surely if we measure tone mapped colour primaries we can find out if the colour is correct and if the saturation is right.
FWIW, my goal is render pixels below diffuse white in such a way that they look exactly the same as they would on a perfect 10,000nits display which needs no tone mapping. It's possible that my processing boosts saturation, but I don't know why it would. I'm wondering if all the HDR movies wouldn't look exactly the same way with the same strong saturation on a perfect 10,000nits display as madVR renders them? I would love to have a way to test this. But I don't know how...

Quote:
Originally Posted by Kris Deering View Post
May I have the exact time stamps for these scenes? I want to check them with my Lumagen just to make sure this isn't a settings issue as I get wonky results if the settings aren't right, so I want to be sure that isn't what is causing these problems.
Would be great if you could double check, just to be safe. I'm also already discussing this with Jim directly, though. We'll get to the bottom of it.
KarlKlammer likes this.
madshi 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