madVR Tool: MadmeasureHDR Optimizer (Measurements, Dynamic Clipping & Target Nits) - Page 3 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 1014Likes
Reply
 
Thread Tools
post #61 of 1208 Old 01-04-2019, 08:09 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,360
Mentioned: 361 Post(s)
Tagged: 0 Thread(s)
Quoted: 5648 Post(s)
Liked: 5978
Quote:
Originally Posted by Soulnight View Post
Nope. That's what algo 1 (50nits) is designed for.

If global picture brightness is very high (meg outside for example) then you will get the same target nits as with the algo 2 (100nits).


They only differentiate themselves for darker scenes.
Algo 1 "50 nits" is the smarter one in that sense.


Basically if you choose 50nits, the formula becomes:

Avg highlights nits + (avg frame nits / avg highlights nits) × avg highlights nits


With:
avg highlights= avg nits of pixels above 100nits

Avg frame nits: basically Fall for the current frame)

For dark scene, the factor(Fall / avg HL) is close to zero. For bright scene like ourside in the meg, equal to 1.

So the target nits algo varies between avg HL nits and 2 times avg HL nits.


With algo 2: it's always 2 times avg HL

And in between a mix of both.


Of course, it's also looking NOT to get higher than the current frame peak. Which could sometimes happen with 2 times avg HL.

Thanks for the explanation. So 50nits will not change all the targets, the way the actual peak changes all the targets with the Manni01 algo?

In that case I'll give 50nits a try, it should work better indeed. Or maybe I should wait for your new algo. Any rough ETA for that?

Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
JVC Macro feature on Vertex/Vertex2/Integral2/Maestro
Manni01 is online now  
Sponsored Links
Advertisement
 
post #62 of 1208 Old 01-04-2019, 08:21 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Manni01 View Post
Thanks for the explanation. So 50nits will not change all the targets, the way the actual peak changes all the targets with the Manni01 algo?

In that case I'll give 50nits a try, it should work better indeed. Or maybe I should wait for your new algo. Any rough ETA for that?
Well I have only one button for "flo" for static Target Nits and dynamic Target Nits.

For static Nits, I recommend not to lie. 100 real nits should be entered as 100.

For dynamic static nits, it's a different story. First it's a different algo (or 2). And it does not use at all the calculated static target nits anyway.

So yes, please, everyone should try dynamic target nits with algo 1 pure with 50 nits. Or algo 2 pure with 100nits. Or a combination of both.



Any other algo will come if needed after that.
Manni01 likes this.
Soulnight is online now  
post #63 of 1208 Old 01-04-2019, 08:26 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,360
Mentioned: 361 Post(s)
Tagged: 0 Thread(s)
Quoted: 5648 Post(s)
Liked: 5978
Quote:
Originally Posted by Soulnight View Post
Well I have only one button for "flo" for static Target Nits and dynamic Target Nits.

For static Nits, I recommend not to lie. 100 real nits should be entered as 100.

For dynamic static nits, it's a different story. First it's a different algo (or 2). And it does not use at all the calculated static target nits anyway.

So yes, please, everyone should try dynamic target nits with algo 1 pure with 50 nits. Or algo 2 pure with 100nits. Or a combination of both.



Any other algo will come if needed after that.
OK, makes sense, will try dynamic algo 1 and will let you know, but it won't be until later today. Thanks!

I guess it would make sense on the GUI to keep the "actual peak nits" when static is selected, and when dynamic is selected, grey this out and offer algo 1 or algo 2 instead.

When you say a combination of both, do you mean selecting a different algo for some titles (for example use 100nits for all titles and remeasure with 50nits for those that have issues, such as the scene in JW2) or do you mean something else?

Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
JVC Macro feature on Vertex/Vertex2/Integral2/Maestro
Manni01 is online now  
Sponsored Links
Advertisement
 
post #64 of 1208 Old 01-04-2019, 08:26 AM
AVS Forum Club Gold
 
Ian_Currie's Avatar
 
Join Date: Nov 2001
Location: Holliston, MA
Posts: 1,734
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 325 Post(s)
Liked: 99
@Soulnight Thanks for answering all my questions!


I spent the day yesterday measuring files and watching the results - and then I noticed your new dynamic version and tried that as well.

Like Dexter, I'm finding that a lot of content gets too darkened by dynamic, but that's a very preliminary impression. I have to play around with it some more. ON titles like The Meg, it's amazing! I am not seeing any brightness pumping which is very impressive.

On a title like Ghost in the Shell, it just increases the target nits way too high for most of the content.

Disclaimers

1) I am still new to using madVR to tonemap (although I've been using it for SDR for at least a year). I've spent the last week doing a lot of reading of the main tonemapping thread and the last 48 hours playing with the software. Prior to this my best results were obtained by a combination of spinning actual discs on my Panny US820, utilizing it's HDR optimization feature to remap any titles above 500 nits to 500, then tonemapping the result to SDR2020 using my Lumagen Pro. This results of this are very good - fantastic shadow detail, punchy image, superb contrast etc. Basically it's the bar I am trying to emulate with madVR (if only because I hate spinning discs).

2) Pathetically, I do NOT know what my peak nits are. I have my projector setup for maximum contrast (iris clamped own etc) and still need to figure that out - so for now I'm assuming 80 nits.


I'm going to be spending my day testing out measured content (my GPU has been going non-stop all night) and switching between Manni's algo and Flo's dynamic one to see and play with the settings. My overall impression with madVR has been mostly VERY positive but there are some titles which don't look as good as I'm used to (so I'm trying to understand why).

I want to say to ALL of you how grateful I am for your contributions. This is fantastic (and fascinating) stuff!

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 #65 of 1208 Old 01-04-2019, 08:32 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Manni01 View Post
OK, makes sense, will try dynamic algo 1 and will let you know, but it won't be until later today. Thanks!

I guess it would make sense on the GUI to keep the "actual peak nits" when static is selected, and when dynamic is selected, grey this out and offer algo 1 or algo 2 instead.

When you say a combination of both, do you mean selecting a different algo for some titles (for example use 100nits for all titles and remeasure with 50nits for those that have issues, such as the scene in JW2) or do you mean something else?
For static there is only one algo. You can see it when you pass your mouse over the recommend static target nits from flo.

For dynamic target nits, there is 2 algo. And it's a smooth blend between the two for any number in between 50 and 100.
At 50 nits, you have algo 1. At 100 its algo 2. At 75nits a 50% blend of the 2. And so on.

Yes, we will present things differently in the future, but I now fully understand Madshi.
Any new field, button, checbox or whatever always costs time
Time you cannot spend on improving algo, or the programm in itself or... watching movies .

Last edited by Soulnight; 01-04-2019 at 08:36 AM.
Soulnight is online now  
post #66 of 1208 Old 01-04-2019, 09:02 AM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
I've given a few problem scenes a try using 50 nits and the results are greatly improved. I'll have to check some brighter scenes as well but I think I prefer this over using 75. I think either way it's a good idea to use average frame nits in your formula. I was playing around with an idea earlier to use something involving percentage of pixels above minimum target nits (as input by the user) but I didn't come up with anything specific, and average frame nits might be easier and better anyway.
Dexter Kane is offline  
post #67 of 1208 Old 01-04-2019, 09:08 AM
Advanced Member
 
grendelrt's Avatar
 
Join Date: Jun 2004
Location: Richmond, VA
Posts: 962
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 578 Post(s)
Liked: 310
Quote:
Originally Posted by Dexter Kane View Post
I've given a few problem scenes a try using 50 nits and the results are greatly improved. I'll have to check some brighter scenes as well but I think I prefer this over using 75. I think either way it's a good idea to use average frame nits in your formula. I was playing around with an idea earlier to use something involving percentage of pixels above minimum target nits (as input by the user) but I didn't come up with anything specific, and average frame nits might be easier and better anyway.
If this is the case, the real display nits field probably needs to be renamed or the values changed. Its going to be really confusing when people are running 100nits actual brightness but setting the real display nits at 50 and the minimum target nits at 100 or 200, it doesn't make much sense for first time users and people not aware of the full definition of the field. This is great stuff though, the automation of not dealing with profiles and being able to change to different profiles inside an individual movie file is amazing.

Last edited by grendelrt; 01-04-2019 at 09:11 AM.
grendelrt is offline  
post #68 of 1208 Old 01-04-2019, 09:30 AM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
That madvr bug seems to be preventing the dynamic tone mapping from working too, which probably doesn't help those dark scenes much.
Dexter Kane is offline  
post #69 of 1208 Old 01-04-2019, 09:40 AM
Member
 
stefanelli73's Avatar
 
Join Date: Jan 2008
Posts: 101
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 27 Post(s)
Liked: 17
I wanted to understand something, with my vpr I have a real peak target of 91nits, at this point in TARGET NITS SELECTION, check the DYNAMIC box and choose FLO, but in ENTER YOUR REAL DISPLAY PEAK NITS I have to enter 50 and in MINIMAL TARGET NITS 91 ?

Last edited by stefanelli73; 01-04-2019 at 10:24 AM.
stefanelli73 is online now  
post #70 of 1208 Old 01-04-2019, 10:14 AM
Advanced Member
 
chros73's Avatar
 
Join Date: Jan 2015
Posts: 592
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 368 Post(s)
Liked: 174
@Soulnight and Anna: thanks for the update!

What settings would you recommend for a TV calibrated at 120 nits in a non-dedicated room + ambient lights? I watched almost (!) every content with your previous algo (v2.4) at 120 nits target
real peak: 50 (?), min: 120, max: 200 (?)

Ryzen 5 2600,Asus Prime b450-Plus,16GB,MSI GTX 1060 Gaming X 6GB(v385.28),Win10 LTSB 1607,MPC-BEx64+LAV+MadVR,Yamaha RX-A870,LG OLED65B8(04.10.25+PC4:4:[email protected]/24/25/29/30/50/59/60Hz)
chros73 is offline  
post #71 of 1208 Old 01-04-2019, 11:26 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by chros73 View Post
@Soulnight and Anna: thanks for the update!

What settings would you recommend for a TV calibrated at 120 nits in a non-dedicated room + ambient lights? I watched almost (!) every content with your previous algo (v2.4) at 120 nits target
real peak: 50 (?), min: 120, max: 200 (?)
Well, if you were happy with 120 target nits, then you love bright picture :-)

The tool will not make it any brighter, but it will adapt it so that the in picture contrast is optimum during the movie. For example the shallow or the meg would be completely overblown with a static target nits of 120nits.

So recommended for your 120nits TV since you love brightness:

Flo + dynamic checked

Algo 1:
Put the real peak at 50

Rolling Avg: 240 frames
Minimum target: 120
Max target: 1000
chros73 likes this.
Soulnight is online now  
post #72 of 1208 Old 01-04-2019, 11:28 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by stefanelli73 View Post
I wanted to understand something, with my vpr I have a real peak target of 91nits, at this point in TARGET NITS SELECTION, check the DYNAMIC box and choose FLO, but in ENTER YOUR REAL DISPLAY PEAK NITS I have to enter 50 and in MINIMAL TARGET NITS 91 ?
Yep. Should work very well like this.
Soulnight is online now  
post #73 of 1208 Old 01-04-2019, 12:15 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,360
Mentioned: 361 Post(s)
Tagged: 0 Thread(s)
Quoted: 5648 Post(s)
Liked: 5978
Quote:
Originally Posted by Soulnight View Post
For static there is only one algo. You can see it when you pass your mouse over the recommend static target nits from flo.

For dynamic target nits, there is 2 algo. And it's a smooth blend between the two for any number in between 50 and 100.
At 50 nits, you have algo 1. At 100 its algo 2. At 75nits a 50% blend of the 2. And so on.

Yes, we will present things differently in the future, but I now fully understand Madshi.
Any new field, button, checbox or whatever always costs time
Time you cannot spend on improving algo, or the programm in itself or... watching movies .
So I did more tests with the following settings:

Real display: 50nits (algo 1 full)
Algo: flo, dynamic
Min: 100nits
Max: 1000nits

The Algo 1 is nice, the targets are brighter than I'm used to, but I couldn't see any downside. The picture is nice and punchy, without any apparent loss of contrast. Maybe a bit less saturation? It definitely improves situations such as the JW2 scene, which looks better without losing anything significant in the highlights (at least that I could see). This should make @Dexter Kane happy. As this setting has nothing to do with real display peak brightness when using dynamic, I understand why he was happier with 75nits. Makes complete sense . I don't mind if the GUI stays like this by the way now that I know how it works, I was only suggesting a change for you to save time in support

Re the 100nits minimum, I wanted to see if there was anything to gain, and there is. The campfire scene in The Revenant, the darker scenes in Blade Runner 2049 or The Meg, they looked nicer without looking too bright, even in my bat loft. But when the content gets brighter and the target remains below 200nits, it does look a lot brighter than I'm used to though, and I'm not sure I need/want this. I haven't seen much unwanted/visible variation either with these extreme settings. Maybe a tad more in some situations, for example in Mad Max 2 or in the attack scene in the Revenant.

These quite extreme settings might cause eye fatigue after a while, I need to do more tests, but it's worth investigating .

Have you tested your algo with 100nits actual peak brightness in a dedicated room (full bat cave, zero nits ambient light), and if you have which settings would you use to keep 0-100nits as stable as possible?
chros73, arcspin and Dexter Kane like this.

Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
JVC Macro feature on Vertex/Vertex2/Integral2/Maestro

Last edited by Manni01; 01-04-2019 at 02:48 PM.
Manni01 is online now  
post #74 of 1208 Old 01-04-2019, 02:33 PM
AVS Forum Club Gold
 
Ian_Currie's Avatar
 
Join Date: Nov 2001
Location: Holliston, MA
Posts: 1,734
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 325 Post(s)
Liked: 99
Quote:
Originally Posted by Manni01 View Post
So I did more tests with the following settings:

Real display: 50nits (algo 1 full)
Algo: flo, dynamic
Min: 100nits
Max: 1000nits

The Algo 1 is nice,
I'm confused, what is Algo 1 vs 2? I don't see any choice of algorithm other than Flo/Manni/Neo-X....

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 #75 of 1208 Old 01-04-2019, 02:47 PM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Ian_Currie View Post
I'm confused, what is Algo 1 vs 2? I don't see any choice of algorithm other than Flo/Manni/Neo-X....
Well I explained it already above quite extensively... ;-)

First, we are now only talking about "Flo" algo for dynamic targets nits.
Then, when you enter "real display nits" = 50, it's what I described above as algo 1.
When you choose " real display nits" = 100, it's algo 2.
And in between it's a blend of the two.

Naming will be changed in the future...
arcspin likes this.

Last edited by Soulnight; 01-04-2019 at 03:30 PM.
Soulnight is online now  
post #76 of 1208 Old 01-04-2019, 02:52 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 8,056
Mentioned: 497 Post(s)
Tagged: 0 Thread(s)
Quoted: 6933 Post(s)
Liked: 6641
Quote:
Originally Posted by Soulnight View Post
Otherwise, and that's for @Javs , you need to go in madVR and select a "master folder for measurement file".
What will happen if you do that is that madVR madmeasureHDR will ITSELF save those measurement files in the said folder.
For this it does not matter what tool you use to generate those measurement file (Manni01 batch, Benrd tool, or ours or even directly the exe from madshi.

As Javs pointed out (I did not know before), the measurements file have been then renamed directly by madVR to include the path to the movie. If you move your movie, it will not work anymore. Same for the measurement files.

Now say you want to optimize your "Measurement folder":
Step 1: you select in madVR your measurement master folder
Step 2: you use one of the tools available and it will create the renamed measurements files in that said folder
Step 3: To optimize them, you need to select in our tool 2 times the path to this folder. Bottom left " measurement file folder" and at the top "paths".
If you keep "same as movie file it will not work because then it looks for mkv.
--> @Javs : you actually did the correct things the first time. I believe you only had access problem because it was on the root. Try again and move the folder somewhere else and select it twice in our programm.
Thanks mate, glad to know I wasnt crazy.

I will try that. Ended up having to do them all one by one, but now you have new features I really hope this will work. I'll try moving the folder to the desktop or something.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #77 of 1208 Old 01-04-2019, 02:53 PM
AVS Forum Club Gold
 
Ian_Currie's Avatar
 
Join Date: Nov 2001
Location: Holliston, MA
Posts: 1,734
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 325 Post(s)
Liked: 99
Quote:
Originally Posted by Soulnight View Post
Well I explained it already above quite extensively... ;-)

First, we are now only talking about "Flo" algo for dynamic targets nits.
Then, when you enter "real display nits" = 50, it's what I described above as algo 1.
When you choose " real display nits" = 100, it's ago 2.
And in between it's a blend of the two.

Naming will be changed in the future...
I'm sorry, I re-read that post a couple times but it didn't sink in. Thanks!

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 #78 of 1208 Old 01-04-2019, 03:47 PM
Member
 
Join Date: Oct 2016
Posts: 15
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 8 Post(s)
Liked: 5
Just for my understanding: What happens if you choose a real display nits below 50 Nits? Let's say 44 nits. Will it still use algo 1 or would you advise against that and choose 50 Nits at lowest for real display nits?
Justkidding is online now  
post #79 of 1208 Old 01-04-2019, 07:25 PM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
@Soulnight

Would it be worth changing the dynamic clipping so that it doesn't clip bellow the minimal target nits?

I've also been watching some movies using the 50 nits setting, and overall I prefer it out of the two algos. But there are still some problem scenes. Die Hard is a good example as most of the movie is under 100nits, in one scene there is only one light which has highlights which drives the target nits up to 800. If I understand your formula right then average highlights is the average of all the pixels above 100 nits? In that case if there are only a small number of pixels above 100 nits but all of them are very bright then the average highlights will be very high. I think it would help to use % of pixels above 100 nits as a weight so that if the percentage if very low it will keep a low target nits but as it goes up the target can get higher.
Manni01, LTD_666 and chros73 like this.

Last edited by Dexter Kane; 01-05-2019 at 01:38 AM.
Dexter Kane is offline  
post #80 of 1208 Old 01-05-2019, 05:17 AM
Advanced Member
 
chros73's Avatar
 
Join Date: Jan 2015
Posts: 592
Mentioned: 12 Post(s)
Tagged: 0 Thread(s)
Quoted: 368 Post(s)
Liked: 174
Quote:
Originally Posted by Soulnight View Post
Looking at the dynamic target nits profile generated with our tool and plotted with @pandm1967 tool, **maybe** 1 min centered rolling avg (1440 frames for 24p content) would be better.
I think the current default is slow enough (maybe it can be even faster): take a look at those graphs, the 240 value reaches higher nits.

@pandm1967 : thanks for the update!

Quote:
Originally Posted by Soulnight View Post
Basically if you choose 50nits, the formula becomes:
Avg highlights nits + (avg frame nits / avg highlights nits) × avg highlights nits
Errr, isn't this equals to: Avg highlights nits + avg frame nits ?

About the Flo algo, I use peak: 50, min: 120
I tried out max 1000, but it was way-way too dark, similarly I wasn't happy about the max 200 setting either.

I readjusted all the files (58 files in 4 minutes ) with max 150 and watched couple of scenes of different movies, now I'm getting closer what I like. My problem is with sunny outside scenes where the image becomes less alive/dynamic/real, it doesn't really affect inside scenes.
Good news is that the min and max limiter is working fine, even with this small interval

Here's an example graph of San Andreas with max 150 setting:
as we can see the values are at the upper limit most of the time and just barely goes down to the min setting (120).

So I guess I'd more like an algo that does the opposite of this: it stays most of the time at the bottom limit (120) and rarely goes up to the upper limit (150).

And now I'm just readjusting the measurement files to have max 135
Attached Thumbnails
Click image for larger version

Name:	capture_025_05012019_121146-01.jpg
Views:	23
Size:	57.9 KB
ID:	2505998  

Ryzen 5 2600,Asus Prime b450-Plus,16GB,MSI GTX 1060 Gaming X 6GB(v385.28),Win10 LTSB 1607,MPC-BEx64+LAV+MadVR,Yamaha RX-A870,LG OLED65B8(04.10.25+PC4:4:[email protected]/24/25/29/30/50/59/60Hz)

Last edited by chros73; 01-05-2019 at 05:21 AM.
chros73 is offline  
post #81 of 1208 Old 01-05-2019, 05:23 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,360
Mentioned: 361 Post(s)
Tagged: 0 Thread(s)
Quoted: 5648 Post(s)
Liked: 5978
Quote:
Originally Posted by chros73 View Post
I think the current default is slow enough (maybe it can be even faster): take a look at those graphs, the 240 value reaches higher nits.

@pandm1967 : thanks for the update!



Errr, isn't this equals to: Avg highlights nits + avg frame nits ?

About the Flo algo, I use peak: 50, min: 120
I tried out max 1000, but it was way-way too dark, similarly I wasn't happy about the max 200 setting either.

I readjusted all the files (58 files in 4 minutes ) with max 150 and watched couple of scenes of different movies, now I'm getting closer what I like. My problem is with sunny outside scenes where the image becomes less alive/dynamic/real, it doesn't really affect inside scenes.
Good news is that the min and max limiter is working fine, even with this small interval

Here's an example graph of San Andreas with max 150 setting:
as we can see the values are at the upper limit most of the time and just barely goes down to the min setting (120).

So I guess I'd more like an algo that does the opposite of this: it stays most of the time at the bottom limit (120) and rarely goes up to the upper limit (150).

And now I'm just readjusting the measurement files to have max 135
Did you try to watch bright films with these settings? You are severely clipping highlights with a 150 max in The Meg, Starship Troopers, Mad Max, Kingsmen - Golden Circle etc.

I'm not sure I see the point of using dynamic with such a small range. Why not use a fixed setting of 135nits and call it a day? Can you even see the difference between 120nits and 135nits target? It's very, very close (at least here).

How much actual peak brightness do you have? If 200nits max is too dim for you, you must have a very low peak brightness. Less than 50nits?
Soulnight likes this.

Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
JVC Macro feature on Vertex/Vertex2/Integral2/Maestro

Last edited by Manni01; 01-05-2019 at 05:38 AM.
Manni01 is online now  
post #82 of 1208 Old 01-05-2019, 05:31 AM
AVS Forum Special Member
 
Neo-XP's Avatar
 
Join Date: Jun 2018
Location: Switzerland
Posts: 1,147
Mentioned: 206 Post(s)
Tagged: 0 Thread(s)
Quoted: 870 Post(s)
Liked: 1212
Quote:
Originally Posted by Soulnight View Post
Looking at the dynamic target nits profile generated with our tool and plotted with @pandm1967 tool, **maybe** 1 min centered rolling avg (1440 frames for 24p content) would be better.
With the default value of 240, I can clearly notice a lot of changes in brightness.

One example: La La Land, from 00:26:05 to 00:26:32.
Manni01 likes this.
Neo-XP is offline  
post #83 of 1208 Old 01-05-2019, 05:45 AM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
Quote:
Originally Posted by Neo-XP View Post
With the default value of 240, I can clearly notice a lot of changes in brightness.

One example: La La Land, from 00:26:05 to 00:26:32.
What value are you using for display nits? When I first tried using 200 I noticed the brightness changes too but since changing to 75 and then 50 I haven't seen any.
Dexter Kane is offline  
post #84 of 1208 Old 01-05-2019, 05:54 AM
Member
 
Lost_62's Avatar
 
Join Date: Apr 2018
Posts: 102
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 65 Post(s)
Liked: 5
@Soulnight.



What advice can you give me for Oled calibrated in a dark room at 100 nits ,Algo 1=50 or Algo 2=100 and (Minimal target nits) ?
Moreover , what a value of nits you would recommend to insert in (Target peak nits) for madVR , keeping in mind that measured brightness it is around a value of 680 nits of brightness.All with Algo FLO.
thanks.
Lost_62 is offline  
post #85 of 1208 Old 01-05-2019, 05:55 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Dexter Kane View Post
@Soulnight

Would it be worth changing the dynamic clipping so that it doesn't clip bellow the minimal target nits?

I've also been watching some movies using the 50 nits setting, and overall I prefer it out of the two algos. But there are still some problem scenes. Die Hard is a good example as most of the movie is under 100nits, in one scene there is only one light which has highlights which drives the target nits up to 800. If I understand your formula right then average highlights is the average of all the pixels above 100 nits? In that case if there are only a small number of pixels above 100 nits but all of them are very bright then the average highlights will be very high. I think it would help to use % of pixels above 100 nits as a weight so that if the percentage if very low it will keep a low target nits but as it goes up the target can get higher.
Yes, that was the plan to scale the avgHL nits with the % of pixels above 100nits.

We need to decide what is "enough" pixels above 100nits to use the full avgHL value.

I was thinking to try with 1% limit to only exclude the crazy cases with dark scenes and very bright tiny spot light driving the avgHL and so the dynamic target nits high for nothing.


So the new "smarter" logic would be like this for both algo 1 and algo 2:
Code:
I would scale avgHL(i) with HLscaling(i)

HLscaling(i) = max[1;%HL(i)/1] with %HL being the sum of the % of the pixels being above 100nits.

Algo 1: (for each frame i)
min[MaxtargetNits;max[MinTargetNits; min[ framepeak(i); avgHL(i)*HLscaling(i) +avgFALL(i) ]]]

Algo 2: (for each frame i)
min[MaxtargetNits;max[MinTargetNits; min[ framepeak(i); 2*avgHL(i)*HLscaling(i)]]]
I will be mostly unavailable in the next days. So please be patient
I won't be able to answer questions much either, so please support each others ;-)
Manni01, neo_2009 and Dexter Kane like this.
Soulnight is online now  
post #86 of 1208 Old 01-05-2019, 06:10 AM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
Quote:
Originally Posted by Soulnight View Post
Yes, that was the plan to scale the avgHL nits with the % of pixels above 100nits.

We need to decide what is "enough" pixels above 100nits to use the full avgHL value.

I was thinking to try with 1% limit to only exclude the crazy cases with dark scenes and very bright tiny spot light driving the avgHL and so the dynamic target nits high for nothing.


So the new "smarter" logic would be like this for both algo 1 and algo 2:
Code:
I would scale avgHL(i) with HLscaling(i)

HLscaling(i) = max[1;%HL(i)/1] with %HL being the sum of the % of the pixels being above 100nits.

Algo 1: (for each frame i)
min[MaxtargetNits;max[MinTargetNits; min[ framepeak(i); avgHL(i)*HLscaling(i) +avgFALL(i) ]]]

Algo 2: (for each frame i)
min[MaxtargetNits;max[MinTargetNits; min[ framepeak(i); 2*avgHL(i)*HLscaling(i)]]]
I will be mostly unavailable in the next days. So please be patient
I won't be able to answer questions much either, so please support each others ;-)
Yeah I think it would only need to be a small percentage, so 1% seems like a good place to start. I'll have a look at some problem scenes and see what the numbers are for them, hopefully it's something consistent.
Manni01 likes this.
Dexter Kane is offline  
post #87 of 1208 Old 01-05-2019, 06:11 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,360
Mentioned: 361 Post(s)
Tagged: 0 Thread(s)
Quoted: 5648 Post(s)
Liked: 5978
Quote:
Originally Posted by Dexter Kane View Post
@Soulnight

Would it be worth changing the dynamic clipping so that it doesn't clip bellow the minimal target nits?

I've also been watching some movies using the 50 nits setting, and overall I prefer it out of the two algos. But there are still some problem scenes. Die Hard is a good example as most of the movie is under 100nits, in one scene there is only one light which has highlights which drives the target nits up to 800. If I understand your formula right then average highlights is the average of all the pixels above 100 nits? In that case if there are only a small number of pixels above 100 nits but all of them are very bright then the average highlights will be very high. I think it would help to use % of pixels above 100 nits as a weight so that if the percentage if very low it will keep a low target nits but as it goes up the target can get higher.
Time code?

Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
JVC Macro feature on Vertex/Vertex2/Integral2/Maestro
Manni01 is online now  
post #88 of 1208 Old 01-05-2019, 06:19 AM
Member
 
Join Date: Nov 2018
Posts: 173
Mentioned: 45 Post(s)
Tagged: 0 Thread(s)
Quoted: 111 Post(s)
Liked: 98
Quote:
Originally Posted by Manni01 View Post
Time code?
It's fairly common in Die Hard, but this is the scene at 00:15:00 that I was talking about specifically

Manni01 likes this.
Dexter Kane is offline  
post #89 of 1208 Old 01-05-2019, 06:22 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Dexter Kane View Post
It's fairly common in Die Hard, but this is the scene at 00:15:00 that I was talking about specifically

Definitely less than 1%.
The other example from John Wick as well.

Could you create one post will all these examples in one. And if you find a new one, always had it there so that everything is in one place.

Time code + screenshot.
Thx.
Manni01 and Dexter Kane like this.
Soulnight is online now  
post #90 of 1208 Old 01-05-2019, 06:24 AM - Thread Starter
AVS Forum Special Member
 
Soulnight's Avatar
 
Join Date: Dec 2011
Location: Germany
Posts: 1,472
Mentioned: 224 Post(s)
Tagged: 0 Thread(s)
Quoted: 1121 Post(s)
Liked: 1823
Quote:
Originally Posted by Neo-XP View Post
With the default value of 240, I can clearly notice a lot of changes in brightness.

One example: La La Land, from 00:26:05 to 00:26:32.
Did you try with algo 1 (50 real peak) ?
Soulnight is online now  
Sponsored Links
Advertisement
 
Reply Home Theater Computers

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