New Lumagen Radiance Pro Series - Page 97 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 3555Likes
Reply
 
Thread Tools
post #2881 of 5223 Old 03-31-2018, 11:20 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Hi,

Currently there is this thread, where improvements to the madVR Tone Mapping are being discussed.
https://www.avsforum.com/forum/24-dig...projector.html

While watching "Life" UHD BD (4000 Nits) I wasn't really happy with IM. So I thought that I also should take a look at a sample scene with some explosions from Mad Max being discussed in the madVR thread.
I attached the resulting pictures of IM compared to a ST.2084-LUT created with LightSpace and a custom curve for my JVC projector created with the Arve tool. I know the overall brightness is slightly different.
The explosions with IM look almost monochrome red, flat and appear to have less contrast.

Also there is another point being discussed in the madVR thread. Some highlights turn blue instead of white. This also happens with IM, but not with LightSpace or Arve. Those are the last three pictures.

I wasn't able to find a combination of parameters (Max display light, Trans, Shape) to improve the result with IM.
It almost feels like revisiting a post I made some time ago
https://www.avsforum.com/forum/37-vid...l#post54473761

The results of IM in the Nit-for-Nit area of the curve are great in my opinion. But I think prior to moving to dynamic tone mapping IM could still use some tweaks for the upper end of the curve. Movies with less than 2000 Nits mastering level are less affected than those above 2000 Nits. But even in Movies <2000 Nits LightSpace-LUTs still have an advantage in the highlights.
Attached Thumbnails
Click image for larger version

Name:	IM1.JPG
Views:	126
Size:	72.1 KB
ID:	2383212   Click image for larger version

Name:	LS1.JPG
Views:	124
Size:	70.3 KB
ID:	2383214   Click image for larger version

Name:	Arve1.JPG
Views:	107
Size:	63.2 KB
ID:	2383216   Click image for larger version

Name:	IM2.JPG
Views:	105
Size:	54.4 KB
ID:	2383218   Click image for larger version

Name:	LS2.JPG
Views:	101
Size:	61.8 KB
ID:	2383220  

Click image for larger version

Name:	Arve2.JPG
Views:	98
Size:	60.7 KB
ID:	2383222   Click image for larger version

Name:	IM3.JPG
Views:	98
Size:	67.6 KB
ID:	2383224   Click image for larger version

Name:	LS3.JPG
Views:	90
Size:	65.4 KB
ID:	2383226   Click image for larger version

Name:	Arve3.JPG
Views:	99
Size:	60.0 KB
ID:	2383228  
bobof likes 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  
Sponsored Links
Advertisement
 
post #2882 of 5223 Old 04-01-2018, 04:05 PM
Advanced Member
 
Roland Janus's Avatar
 
Join Date: Mar 2005
Location: Switzerland
Posts: 539
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 133 Post(s)
Liked: 33
I'm wondering how can -4 to +4 be within a 10 stripe pattern?

-4 to +4 is nine steps, so which one is the 0?

-Roland
Attached Thumbnails
Click image for larger version

Name:	Capture.PNG
Views:	53
Size:	12.2 KB
ID:	2383732  
Roland Janus is offline  
post #2883 of 5223 Old 04-02-2018, 12:54 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by KarlKlammer View Post
I attached the resulting pictures of IM compared to a ST.2084-LUT created with LightSpace and a custom curve for my JVC projector created with the Arve tool. I know the overall brightness is slightly different.
The explosions with IM look almost monochrome red, flat and appear to have less contrast.

Also there is another point being discussed in the madVR thread. Some highlights turn blue instead of white. This also happens with IM, but not with LightSpace or Arve. Those are the last three pictures.
I'm just getting set up with my new Lumagen Pro and found your post useful. I found the car blue highlight scene at the beginning of Mad Max Fury Road but I'm not familiar with the movie - do you have timestamps for the explosions too? I'd like to play around with them myself.

I have Lightspace HTP. For now I'd just loaded in the plain 2.4 SDR 2020 LUT for my projector and trying Intensity Mapping, but of course I see the same blue highlight issue on that car (be nice to see if this can be improved in a future version of IM). Out of interest, when you're making dedicated LUT for HDR conversion in Lightspace, how many LUTs do you have for HDR? Just one or several? I've previously hacked around with my old Lumagen 2143 doing HDR conversion in LUT so I know my way around Lightspace for that.
bobof is online now  
Sponsored Links
Advertisement
 
post #2884 of 5223 Old 04-02-2018, 01:44 PM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Quote:
Originally Posted by KarlKlammer View Post
Hi,

Currently there is this thread, where improvements to the madVR Tone Mapping are being discussed.
https://www.avsforum.com/forum/24-dig...projector.html

While watching "Life" UHD BD (4000 Nits) I wasn't really happy with IM. So I thought that I also should take a look at a sample scene with some explosions from Mad Max being discussed in the madVR thread.
I attached the resulting pictures of IM compared to a ST.2084-LUT created with LightSpace and a custom curve for my JVC projector created with the Arve tool. I know the overall brightness is slightly different.
The explosions with IM look almost monochrome red, flat and appear to have less contrast.

Also there is another point being discussed in the madVR thread. Some highlights turn blue instead of white. This also happens with IM, but not with LightSpace or Arve. Those are the last three pictures.

I wasn't able to find a combination of parameters (Max display light, Trans, Shape) to improve the result with IM.
It almost feels like revisiting a post I made some time ago
https://www.avsforum.com/forum/37-vid...l#post54473761

The results of IM in the Nit-for-Nit area of the curve are great in my opinion. But I think prior to moving to dynamic tone mapping IM could still use some tweaks for the upper end of the curve. Movies with less than 2000 Nits mastering level are less affected than those above 2000 Nits. But even in Movies <2000 Nits LightSpace-LUTs still have an advantage in the highlights.

Having looked at Mad Max last night in this very scene I can say first hand that my results look NOTHING like your IM sample and are far more in line with what was shown with the Arve and LS examples. I think this is a setup issue on your part, not a limitation of IM.

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2885 of 5223 Old 04-02-2018, 02:04 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by bobof View Post
I'm just getting set up with my new Lumagen Pro and found your post useful. I found the car blue highlight scene at the beginning of Mad Max Fury Road but I'm not familiar with the movie - do you have timestamps for the explosions too? I'd like to play around with them myself.

I have Lightspace HTP. For now I'd just loaded in the plain 2.4 SDR 2020 LUT for my projector and trying Intensity Mapping, but of course I see the same blue highlight issue on that car (be nice to see if this can be improved in a future version of IM). Out of interest, when you're making dedicated LUT for HDR conversion in Lightspace, how many LUTs do you have for HDR? Just one or several? I've previously hacked around with my old Lumagen 2143 doing HDR conversion in LUT so I know my way around Lightspace for that.
You can find plenty of those scenes between ~0:28:00 and 0:30:00. My first picture is at 0:28:29, and the second 0:28:37.
To cross-check if the blue highlights of those chome elements or the rather dull explosions are caused by a failed LUT I uploaded a plain cube without actual calibration data. The blue highlights remain even then. The same goes for the explosions.
In the madVR thread it is suspected that the blue highlights are an enconding issue. That alone doesn't explain why they turn blue in some tone mapping solutions but not in others.

After the changes last year I switched from LightSpace ST.2084-LUTs to IM, because I found it more flexible without the need to always connect my PC to do some changes, even though I liked some aspect of the LightSpace tone mapping better.
Before that I had one ST.2084-LUT for 1000 nit mastered discs and one for 4000 nit mastered discs. Currently I have one CMS slot reserved for IM (PJ in low lamp mode), one slot for an ST.2084-LUT (4000 nits), one slot for some Arve curves ("UHD Rec.2020", Gamma 2.2) and one slot for IM (PJ in high lamp mode).

I am absolutely certain that IM could be improved and it is "only" a mathematical issue.
bobof likes this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-02-2018 at 02:52 PM.
KarlKlammer is offline  
post #2886 of 5223 Old 04-02-2018, 02:30 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by Kris Deering View Post
Having looked at Mad Max last night in this very scene I can say first hand that my results look NOTHING like your IM sample and are far more in line with what was shown with the Arve and LS examples. I think this is a setup issue on your part, not a limitation of IM.
That could certainly be the case, though I already heard from another user that for those scenes he also liked his LightSpace LUTs better than IM.
Maybe bobof can share his thoughts when he has completed his setup.

Edit:
The last time I checked my calibration for IM my Gamut looked like this
Attached Thumbnails
Click image for larger version

Name:	JVC_A_SDR2020_Sweep75_20180226.JPG
Views:	73
Size:	53.6 KB
ID:	2384066  
bobof likes this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-02-2018 at 02:43 PM.
KarlKlammer is offline  
post #2887 of 5223 Old 04-02-2018, 02:38 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by Kris Deering View Post
Having looked at Mad Max last night in this very scene I can say first hand that my results look NOTHING like your IM sample and are far more in line with what was shown with the Arve and LS examples. I think this is a setup issue on your part, not a limitation of IM.
Are you referring to all the Mad Max images, or just the explosions? I've not looked at the explosions yet, but my car highlight also looks blue under IM very much like the image above, so if there is something specific which prevents that happening it would be great to know . How do you have your unit set up?
bobof is online now  
post #2888 of 5223 Old 04-02-2018, 03:26 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by KarlKlammer View Post
Maybe bobof can share his thoughts when he has completed his setup.
For what it is worth I also see what is best described as a lack of yellow in the explosions in IM, which seem to come back if I create ST2084 LUT in LS. I have no idea which is "correct" though!

I'm sure the hard border itself on the metal highlights of the car is in the encoding, as you say the more interesting thing is the colour it takes on.

My setup is very far from complete though. At the moment I just have the Pro hacked into my lounge system with an old JVC X30 while I wait for the dedicated room to be done which will get an X7900. I used an old 17^3 profile to create the UHD2020 LUT for the basis of IM.
KarlKlammer likes this.
bobof is online now  
post #2889 of 5223 Old 04-02-2018, 05:12 PM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Quote:
Originally Posted by KarlKlammer View Post
That could certainly be the case, though I already heard from another user that for those scenes he also liked his LightSpace LUTs better than IM.
Maybe bobof can share his thoughts when he has completed his setup.

Edit:
The last time I checked my calibration for IM my Gamut looked like this
Your gamut isn't the issue, it is the tone map. IM isn't shrinking saturation, I've already confirmed that last week when people were wondering if IM shrinks like the Oppo tone mapping does.

I would expect different tone mapping from solution to solution though since each is doing it differently. With IM there is a LOT of variability in what can be achieved, so the values you are using for the example pictures is not ideal.

The blue in the Mad Max clip is seen with IM regardless. I am talking about the explosions. The settings there are way off.

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2890 of 5223 Old 04-02-2018, 05:16 PM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Quote:
Originally Posted by bobof View Post
Are you referring to all the Mad Max images, or just the explosions? I've not looked at the explosions yet, but my car highlight also looks blue under IM very much like the image above, so if there is something specific which prevents that happening it would be great to know . How do you have your unit set up?
Again, the issue here is not gamut. IM does not hurt saturation or shrink the gamut of the projector. I don't even use the 3D LUT for 2020 since OOTB my projector's tracking is nearly spot on when measured in CalMan (JVC RS600 with BT2020 profile). What is wrong is the tone map. Since this is a completely subjective setting it would be easy to have them set wrong and get results that would not look as good as the Arve tool or other solutions. Again, I had looked at this very scene last night and it looks NOTHING like the example shown here for IM. I've compared the Arve tool MANY times to IM, including the popular curves from Javs and Manni (including custom curves created by them specifically for my setup) and every time I've found IM to work better across all the different films/clips I use for comparison (including the Mad Max scene shown above).

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2891 of 5223 Old 04-02-2018, 10:18 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by Kris Deering View Post
Your gamut isn't the issue, it is the tone map. IM isn't shrinking saturation, I've already confirmed that last week when people were wondering if IM shrinks like the Oppo tone mapping does.

I would expect different tone mapping from solution to solution though since each is doing it differently. With IM there is a LOT of variability in what can be achieved, so the values you are using for the example pictures is not ideal.

The blue in the Mad Max clip is seen with IM regardless. I am talking about the explosions. The settings there are way off.
OK, so these are the settings last used for 116 Nits measured peak white
Color Space: SDR2020
Display max Light: 500
Gamma to 3D LUT: SDR
Ratio: 0
Shape: 5
Trans: 12
Gamma: 0
Black: 0

My initial Setting for Shape was 7 and 14 for Trans. But that really doesn't matter as I tried numerous combinations. There is always way to little yellow visible in those explosions.

I don't think this is a sign for a shrunken Gamut. Why would yellow turn red, if the Gamut shrinks?

So in your setup you use the BT.2020 color profile of the JVC?
I use "Reference" with Gamma 2.4. And I know from Gordon Fraser that he uses a complete turned off CMS on JVCs. The LUT in my setup then converts the Gamut with DCI-P3 saturation points to BT.2020, as you can see in my chart.

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 #2892 of 5223 Old 04-03-2018, 12:38 AM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by Kris Deering View Post
Again, the issue here is not gamut. IM does not hurt saturation or shrink the gamut of the projector. I don't even use the 3D LUT for 2020 since OOTB my projector's tracking is nearly spot on when measured in CalMan (JVC RS600 with BT2020 profile). What is wrong is the tone map. Since this is a completely subjective setting it would be easy to have them set wrong and get results that would not look as good as the Arve tool or other solutions. Again, I had looked at this very scene last night and it looks NOTHING like the example shown here for IM. I've compared the Arve tool MANY times to IM, including the popular curves from Javs and Manni (including custom curves created by them specifically for my setup) and every time I've found IM to work better across all the different films/clips I use for comparison (including the Mad Max scene shown above).
Thanks for the reply Kris. It sounds like I have my setup configured a bit more like Karl than you - output is set to REC709SDR, with a LUT to map the colourspace I have available in the JVC X30 as best as possible to SDR REC2020, with IM set to SDR gamma in the LUT. In my case this setting at the moment is out of necessity due to the old PJ, but I was thinking of running like this later with the X7900 to avoid the noise from the filter banging in and out and get a bit more light. I find the levers do only a little to adjust the yellow sections. I also experimented with max light from 100-500 and I find IM looks better towards the bottom of the range to me.

As I said, I have no idea what look is "right" though for these images, I'm merely seeing a similar difference between LS dedicated LUT and IM to what the images above show. And there is, as you point out, a lot of subjectivity here. You mention is is possible (of course) to set the IM parameters "wrong" - can I ask what is your procedure for setting them "right"?

Is your output set to SDR2020 or HDR2020? And your gamma in 3DLUT set to SDR, Auto or HDR? Could you share how you have your IM configured?

Thanks

Last edited by bobof; 04-03-2018 at 02:55 AM.
bobof is online now  
post #2893 of 5223 Old 04-03-2018, 01:21 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by bobof View Post
I also experimented with max light from 100-500 and I find IM looks better towards the bottom of the range to me.
Can you tell your measured peak white?
What is the multiplier you used for the ST.2084-LUT?

I don't want to go below 500 Nits Display Max Light, because my image is bright enough with this.
By the way, when I increase Display Max Light, the scenes only get darker but colors stay the same.

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 #2894 of 5223 Old 04-03-2018, 03:03 AM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by KarlKlammer View Post
Can you tell your measured peak white?
What is the multiplier you used for the ST.2084-LUT?

I don't want to go below 500 Nits Display Max Light, because my image is bright enough with this.
By the way, when I increase Display Max Light, the scenes only get darker but colors stay the same.
The profile I used for the ST2084 LUT was only I think around 65cdm2 max measured as I didn't do a new profile with aperture fully open; if I recall I set the multiplier to 3x.
My preference is for the Display Max Light to be in the 200-300 range, but my current PJ in low lamp mode with aperture fully open only reaches 80cdm2, so I guess I need different settings to have a bright enough image to be useful. Your peak white is much brighter than mine.

As I say I'm only hacking around with this quickly at the moment, hope to play some more when I have time and new room / projector.
bobof is online now  
post #2895 of 5223 Old 04-03-2018, 03:13 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by bobof View Post
The profile I used for the ST2084 LUT was only I think around 65cdm2 max measured as I didn't do a new profile with aperture fully open; if I recall I set the multiplier to 3x.
My preference is for the Display Max Light to be in the 200-300 range, but my current PJ in low lamp mode with aperture fully open only reaches 80cdm2, so I guess I need different settings to have a bright enough image to be useful. Your peak white is much brighter than mine.

As I say I'm only hacking around with this quickly at the moment, hope to play some more when I have time and new room / projector.
OK, then it makes sense to go way below 500 nits for Display Max light.
My multiplier for a ST.2084-LUT is 5x. So it is about the same range as Display Max Light.
bobof likes this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-03-2018 at 03:56 AM.
KarlKlammer is offline  
post #2896 of 5223 Old 04-03-2018, 06:30 AM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by KarlKlammer View Post
OK, then it makes sense to go way below 500 nits for Display Max light.
My multiplier for a ST.2084-LUT is 5x. So it is about the same range as Display Max Light.
I started at 5 but always found that too dark.

Sent from my E5823 using Tapatalk
bobof is online now  
post #2897 of 5223 Old 04-03-2018, 07:57 AM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Looked at this more last night. Great work here guys. So I can confirm your findings. It is interesting, if you watch that scene in its entirety, the amount of "yellow flames" varies. So if I hadn't paused on exactly the same frame as you, I wouldn't have noticed this as much. But I can confirm the findings when comparing directly to the Arve curves I have loaded.


I then spent nearly two hours on the phone with Jim as we tried to figure it out. The only way to get the yellow back in the image was to raise the white level in the Radiance menu. Jim thinks it may be a clipping issue. I know they are busy on a lot of things so not sure if or when they'll have time to work on this, but he is aware and the gears are turning in his head!


Tone mapping is tough. Without side by side comparisons it is hard to find instances like this. And for ones that I find where the Arve tool works, I find others where IM works better (obviously subjectively).
Spizz, alex_t, Kelvin1965S and 2 others like this.

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2898 of 5223 Old 04-03-2018, 08:16 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by Kris Deering View Post
Looked at this more last night. Great work here guys. So I can confirm your findings. It is interesting, if you watch that scene in its entirety, the amount of "yellow flames" varies. So if I hadn't paused on exactly the same frame as you, I wouldn't have noticed this as much. But I can confirm the findings when comparing directly to the Arve curves I have loaded.


I then spent nearly two hours on the phone with Jim as we tried to figure it out. The only way to get the yellow back in the image was to raise the white level in the Radiance menu. Jim thinks it may be a clipping issue. I know they are busy on a lot of things so not sure if or when they'll have time to work on this, but he is aware and the gears are turning in his head!


Tone mapping is tough. Without side by side comparisons it is hard to find instances like this. And for ones that I find where the Arve tool works, I find others where IM works better (obviously subjectively).
Yes Kris, it is a little bit tricky to find, even more if you don't pause playback at the right moment.
After your comments I was beginning to question my sanity. But thanks for confirming it and contacting Jim!

I don't know if it is helpful for IM. But this is how it seems to be handled by madVR:
Quote:
Basically I'm converting YCbCr to RGB, then I'm simply hard clipping every RGB value outside of 16-235 to 16-235.
https://www.avsforum.com/forum/24-dig...l#post55928826

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 #2899 of 5223 Old 04-03-2018, 09:32 AM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Quote:
Originally Posted by KarlKlammer View Post
Yes Kris, it is a little bit tricky to find, even more if you don't pause playback at the right moment.
After your comments I was beginning to question my sanity. But thanks for confirming it and contacting Jim!

I don't know if it is helpful for IM. But this is how it seems to be handled by madVR:
https://www.avsforum.com/forum/24-dig...l#post55928826

So it sounds like Madshi is in the same wheelhouse as Jim, that the way IM is handling it is CORRECT. We looked at the raw data last night and Jim was saying nearly every single thing that Madshi says in that thread about how having yellow in that scene makes no sense based on the input values. It is because the Arve tool is doing something illegal that you get the result you see (which I agree subjectively looks better, whether it is right or wrong). But it does bring up the same issue Madshi mentioned and Jim mentioned last night, do you want to actually do the WRONG thing for colors simply so this ONE scene can look subjectively better. And what happens when that implementation ends up making other scenes look bad because of it?

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2900 of 5223 Old 04-03-2018, 10:00 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by Kris Deering View Post
So it sounds like Madshi is in the same wheelhouse as Jim, that the way IM is handling it is CORRECT. We looked at the raw data last night and Jim was saying nearly every single thing that Madshi says in that thread about how having yellow in that scene makes no sense based on the input values. It is because the Arve tool is doing something illegal that you get the result you see (which I agree subjectively looks better, whether it is right or wrong). But it does bring up the same issue Madshi mentioned and Jim mentioned last night, do you want to actually do the WRONG thing for colors simply so this ONE scene can look subjectively better. And what happens when that implementation ends up making other scenes look bad because of it?
Yes, madshi seems to be in a similar position. But if I remember it correctly, the main issue with madVR was desaturation. He and Jim certainly have valid points with the scientific approach.
But unless someone in the industry says clearly how it is supposed to look like or how the tools used to master it are working, this will always be a guessing game.

I just watched the scenes on my Samsung KS7090 without Radiance or other tools involved. The result is similar to LightSpace ST.2084 or an Arve curve.

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 #2901 of 5223 Old 04-03-2018, 02:55 PM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,694
Mentioned: 188 Post(s)
Tagged: 0 Thread(s)
Quoted: 3830 Post(s)
Liked: 6525
Quote:
Originally Posted by KarlKlammer View Post
Yes, madshi seems to be in a similar position. But if I remember it correctly, the main issue with madVR was desaturation. He and Jim certainly have valid points with the scientific approach.
But unless someone in the industry says clearly how it is supposed to look like or how the tools used to master it are working, this will always be a guessing game.

I just watched the scenes on my Samsung KS7090 without Radiance or other tools involved. The result is similar to LightSpace ST.2084 or an Arve curve.
But there is no guess here. As Madshi clearly points out in the thread, while it may have looked like that for the person mastering, it would only look that way if the display had improper tone mapping, which definitely could be the case. I think a lot of people always assume that the powers that be know what is right or wrong, but the more I've been around "experts" and industry types, the more I find that is not always the case.


The problem here though lies in the fact that this movie (Mad Max) was mastered on a Pulsar, so tone mapping should never have been done in the first place, BUT, a lot of the data on the disc indicates levels that would have exceeded that monitor AND are illegal in just about every case. MadVR and the Lumagen IM convert the levels to LEGAL liner light levels, whereas other solutions are using ILLEGAL levels that should never be there in the first place (again, because the tone mapping is doing things VERY wrong, as Madshi points out). In this very particular case it gives subjectively more pleasing results (for a scene that last about a minute overall) but at the expense of how much with other titles?


Ultimately it looks like I was right when I talked to Jim about all this last night, Mad Max is a horrible film to use for evaluation of tone mapping as it CLEARLY has authoring issues that wreak havoc. I also think it is a bad one because ultimately all these solutions are being played around with for "subjective" opinions on what looks better than the other since there is no standard and this is a film that is so heavy handed in post production color work that trying to determine what looks right is nearly impossible!

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is online now  
post #2902 of 5223 Old 04-04-2018, 12:31 AM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
With IM, LightSpace and Arve I now have three tone mapping solutions at hand. I don't question color science in general. But as a consumer, I have to say that I'm going to use the solution that looks best to me. And currently in this case or with other movies with MaxCLL >4000 nits it is likely not going to be IM.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-04-2018 at 07:00 AM.
KarlKlammer is offline  
post #2903 of 5223 Old 04-04-2018, 11:08 AM
AVS Forum Club Gold
 
Ian_Currie's Avatar
 
Join Date: Nov 2001
Location: Holliston, MA
Posts: 1,705
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 294 Post(s)
Liked: 88
Quote:
Originally Posted by KarlKlammer View Post
With IM, LightSpace and Arve I now have three tone mapping solutions at hand. I don't question color science in general. But as a consumer, I have to say that I'm going to use the solution that looks best to me. And currently in this case or with other movies with MaxCLL >4000 nits it is likely not going to be IM.
I agree. I'm using a combination of techniques to get the best picture quality. My preference is to NOT convert HDR2020 to SDR2020 but to stay in HDR2020 mode. When I convert I get lots of blotchy effects (easily visible on faces) that no amount of adjusting shape and trans can fix without compromising the picture in other areas) or loss of color saturation (a look that is more SDR than HDR) .

In my case, I alternate between a couple Arve curves ... or the HDR gamma on my JVC combined with Lumagen IM that does not convert to SDR. Depending on the movie I may go to high laser (i.e. highest light output) but the results are spectacular.

I wish there was more control over the IM when staying in HDR2020 mode but obviously IM wasn't intended to be used this way.

JVC-RS4500 4k projector, Lumagen Pro, 138" 2.35:1 ST130
Panasonic DMP-UB820 & i7 PC w/1080ti running Win10/Kodi DSPlayer/madVR
Classe SSP-800, Bryston amplification, Wilson Audio speakers (7.2) + Buttkickers.
Ian_Currie is offline  
post #2904 of 5223 Old 04-04-2018, 12:46 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Since it was established that Mad Max is a bad movie to evaluate tone mapping, originally "Life" (6414 nits MaxCLL) was the reason I started comparing IM to other tone mapping solutions again. These are two pictures of IM and LightSpace.
I know what my projector is capable of regarding contrast. And because of that I have a hard time when the image seems to lack contrast. I wasn't able to find settings that reduced the general "fog" without compromising other aspects of the image.
Attached Thumbnails
Click image for larger version

Name:	IM_Life.JPG
Views:	100
Size:	141.3 KB
ID:	2384880   Click image for larger version

Name:	LS_Life.JPG
Views:	96
Size:	147.9 KB
ID:	2384882  

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-04-2018 at 12:57 PM.
KarlKlammer is offline  
post #2905 of 5223 Old 04-04-2018, 01:42 PM
AVS Forum Club Gold
 
Ian_Currie's Avatar
 
Join Date: Nov 2001
Location: Holliston, MA
Posts: 1,705
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 294 Post(s)
Liked: 88
Quote:
Originally Posted by KarlKlammer View Post
Since it was established that Mad Max is a bad movie to evaluate tone mapping, originally "Life" (6414 nits MaxCLL) was the reason I started comparing IM to other tone mapping solutions again. These are two pictures of IM and LightSpace.
I know what my projector is capable of regarding contrast. And because of that I have a hard time when the image seems to lack contrast. I wasn't able to find settings that reduced the general "fog" without compromising other aspects of the image.
Good example, thanks for sharing.

JVC-RS4500 4k projector, Lumagen Pro, 138" 2.35:1 ST130
Panasonic DMP-UB820 & i7 PC w/1080ti running Win10/Kodi DSPlayer/madVR
Classe SSP-800, Bryston amplification, Wilson Audio speakers (7.2) + Buttkickers.
Ian_Currie is offline  
post #2906 of 5223 Old 04-04-2018, 02:44 PM
aka jfinnie
 
Join Date: Aug 2015
Location: Norwich, UK
Posts: 3,005
Mentioned: 49 Post(s)
Tagged: 0 Thread(s)
Quoted: 2430 Post(s)
Liked: 1563
Quote:
Originally Posted by KarlKlammer View Post
Since it was established that Mad Max is a bad movie to evaluate tone mapping, originally "Life" (6414 nits MaxCLL) was the reason I started comparing IM to other tone mapping solutions again. These are two pictures of IM and LightSpace.
I know what my projector is capable of regarding contrast. And because of that I have a hard time when the image seems to lack contrast. I wasn't able to find settings that reduced the general "fog" without compromising other aspects of the image.
It is a really difficult one though isn't it as you could equally look at the LS image and say that in the shadow regions it seems some of the detail is starting to be crushed.

I think that whether or not the Mad Max yellow explosions are right or wrong, it would be nice to at least have IM levers to be able to pull to get to a similar point to what the LS/ Arve examples end up.

But whatever you think of the individual images and right vs wrong, better vs worse, contrast vs shadows, to me it absolutely reaffirms how important it is to be able to make these kind of adjustments and just how glad I am that I decided to go down the Radiance route again with HDR. MadVR isn't a good option for me as much of the content I watch is streamed via boxes with no HDR controls at all.

It is also great that we're able to have sensible conversation about them here - I'm sure it can only get better

Sent from my E5823 using Tapatalk
Kelvin1965S likes this.
bobof is online now  
post #2907 of 5223 Old 04-04-2018, 10:32 PM
AVS Forum Special Member
 
westmd's Avatar
 
Join Date: Jan 2010
Posts: 1,309
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 977 Post(s)
Liked: 405
A need a bit of help from the community here. I was going for a Lumagen Pro and was about to order one when I got some budget cuts due to renovations that needs to be doing in the housr. This also means less time and as I am a complete color calibration newbie I would most likely have to invest a lot of time to calibrate the Lumagen. So in the end I got a direct calibration for my JVC X7000 and am more then happy with the result.
In the meantime I got an offer for a trade-in Lumagen Pro for quite a good price. As restrictions still apply (time/money) I would go without software/probe and just go with the Lumagen for scaling and HDR intensity mapping (I currently use the Oppo 203 for that).
Do you think it is worth getting the Lumagen for this purposes and will especially Intensity mapping be as good without creation of a calibration 3D LUT?
Thanks for your advice
westmd is offline  
post #2908 of 5223 Old 04-04-2018, 10:34 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Quote:
Originally Posted by bobof View Post
It is a really difficult one though isn't it as you could equally look at the LS image and say that in the shadow regions it seems some of the detail is starting to be crushed.
Yes, and this is a point where I like the flexibility of IM. With a few adjustments you can get every shadow detail that is there. Some shadow details might be missing because I had the same aperture and exposure time on both pictures (focused on the brighter parts). Even with Arve you don't miss shadow details. But I haven't found a solution with LS yet. If you start to use the Radiance Gamma-EQ you can easily run into banding or colour shifts.

Quote:
I think that whether or not the Mad Max yellow explosions are right or wrong, it would be nice to at least have IM levers to be able to pull to get to a similar point to what the LS/ Arve examples end up.

But whatever you think of the individual images and right vs wrong, better vs worse, contrast vs shadows, to me it absolutely reaffirms how important it is to be able to make these kind of adjustments and just how glad I am that I decided to go down the Radiance route again with HDR. MadVR isn't a good option for me as much of the content I watch is streamed via boxes with no HDR controls at all.

It is also great that we're able to have sensible conversation about them here - I'm sure it can only get better
After pointing out differences between IM and LS last year and using The Magnificent Seven as an example, the changes made definitely improved the result. But there are still moments when I wish IM would be better.
It is not my intention to endlessly complain about IM or the Radiance. I only want to point out that there is still room for improvement and hope it would be made. I really don't regret having bought the Radiance. For me madVR also isn't an option. Apart from the fact that it isn't a plug-in-videoprocessor for serveral sources, I don't want to rely on a PC with Windows for a relaxed evening in my home cinema.
Kelvin1965S and bobof like this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-04-2018 at 11:30 PM.
KarlKlammer is offline  
post #2909 of 5223 Old 04-06-2018, 11:51 PM
Senior Member
 
KarlKlammer's Avatar
 
Join Date: Apr 2010
Location: Germany
Posts: 306
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 256 Post(s)
Liked: 247
Has anybody ever wondered, what it would be like to be able to connect to the Radiance via (wireless) network instead of RS232/USB?
So, there are those things called "Serial Device Servers". I just got one from MOXA with two ports:
https://www.moxa.com/product/NPort_5200A.htm

I made some quick tests with my laptop only connected to wifi and the i1 Display Pro:
- downloading the configuration with the Config-Tool -> OK
- manually selecting Patches in LightSpace -> OK
- LightSpace Quick Profiles -> OK
- LightSpace LUT upload -> OK
- Chromapure with connection to the Radiance as pattern generator -> OK
- Radiance-GUI by Alex_t (extremely helpful tool by the way!) -> OK
- firmware upload -> OK

So far it looks very promising. I expected more problems with my first tests. But long-term stability has still to be proved.

Edit:
- LightSpace Characterization (17^3 cube with drift compensation every 30 patches, 79 minutes duration) -> OK
Gordon Fraser, mikela, jrp and 3 others like this.

Projection: JVC DLA-NX9
VP/Calibration: Lumagen Radiance Pro, LightSpace CMS, x-rite i1 Pro 2, x-rite i1 Display 3

Last edited by KarlKlammer; 04-08-2018 at 01:59 AM.
KarlKlammer is offline  
post #2910 of 5223 Old 04-07-2018, 02:27 AM
Senior Member
 
alex_t's Avatar
 
Join Date: May 2006
Location: France
Posts: 421
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 130
Quote:
Originally Posted by KarlKlammer View Post
Has anybody ever wondered, what it would be like to be able to connect to the Radiance via (wireless) network instead of RS232/USB?
So, there are those things called "Serial Device Servers". I just got one from MOXA with two ports:
https://www.moxa.com/product/NPort_5200A.htm

I made some quick tests with my laptop only connected to wifi and the i1 Display Pro:
- downloading the configuration with the Config-Tool -> OK
- manually selecting Patches in LightSpace -> OK
- LightSpace Quick Profiles -> OK
- LightSpace LUT upload -> OK
- Chromapure with connection to the Radiance as pattern generator -> OK
- Radiance-GUI by Alex_t (extremely helpful tool by the way!) -> OK
- firmware upload -> OK

So far it looks very promising. I expected more problems with my first tests. But long-term stability has still to be proved.
Hello.

Yes I have already wondered if such solution would work, thank you for your return of experience.

Actually I would like to add in Radiance-GUI the possibility to set up HDR intensity mapping (shape, transition, gamma, black) according to a data base built with users's experience. For example, Radiance GUI would allow to select a film in this data base and Radiance-GUI would inject the IM settings. Of course those settings depend of Max display light value but I believe this feature could be helpful for users as starting point to fine tune IM.

Radiance-GUI would be also used for collecting statistics and other useful information about HDR and the film watched with the possibility to send automatically those information to me for feeding the data base.

With this HW solution, connecting Radiance-GUI to Radiance becomes handier and this kind of feature would make sense.
KarlKlammer likes this.

Last edited by alex_t; 04-07-2018 at 02:37 AM.
alex_t is offline  
Sponsored Links
Advertisement
 
Reply Video Processors



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