Improving Madvr HDR to SDR mapping for projector - Page 24 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 4800Likes
Reply
 
Thread Tools
post #691 of 6092 Old 04-25-2018, 04:33 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by fhoech View Post
BT.2020 on the projector side surely is P3-ish within 2020, and this then means that it could actually be the projector that is clipping, not madVR. When madVR is set to BT.2020, it expects the display to have the full BT.2020 gamut, I think (probably there needs to be an option for P3 within 2020 container on the calibration tab?).
Can you send/share the CalMAN LUT?
This is entirely irrelevant.

Of course the display only covers P3 (about 97% of it) but it uses a BT2020 profile with BT2020 saturations for my SDR BT2020 baseline. P3 isn't relevant with consumer content from a calibration point of view, so just forget about it.

The content itself (at least the content I use for my test) also doesn't go further than P3, as confirmed by MadVR and by the Vertex.

We're only using BT2020 as a container, to encode P3 content, and then to display this P3 content on the projector.

In any case, there is no clipping with dumb, or using a higher target nits to reduce the highlights compression.

If it was the projector clipping, it would be clipping irrespective of the mode (dumb or complex) and it would be clipping at 600nits as well at 500nits. It would also be clipping in passthrough, which isn't the case.

This suggests that it's the highlights that are clipping (due to a different way to handle them, psossibly because they are redder), not the content at the edge of the gamut. It also causes contouring.

Again, this content (up to 4000nits and above) should be displayed with a target nits of 600nits in order not to clip. when the target is 600nits, neither dumb nor scientific clip. But if I lower the target nits to 500, dumb still doesn't clip and scientific does.

This simply allows me to use a better compromise for 1000nits titles (500nits target) until MadVR handles content in a more adaptive way and doesn't only deal with highlights with bright titles but also improves the low end with dimmer titles.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
Sponsored Links
Advertisement
 
post #692 of 6092 Old 04-25-2018, 04:50 AM
AVS Forum Special Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 1,047
Mentioned: 72 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 378
Quote:
Originally Posted by Manni01 View Post
This is entirely irrelevant.
I fail to see why.

Quote:
Originally Posted by Manni01 View Post
The content itself (at least the content I use for my test) also doesn't go further than P3, as confirmed by MadVR and by the Vertex.
After the BT2390 roll-off and tonemapping is applied, and the destination gamut is 2020 in madVR, the tonemapped values may well exceed the (roughly) effective P3 gamut of the projector (unless the tonemapping already makes sure that the tonemapped values can never exceed P3, which would seem odd).

Quote:
Originally Posted by Manni01 View Post
It would also be clipping in passthrough, which isn't the case.
Well, in passthrough mode the original encoded values won't exceed P3 (on the signal side) because that's usually how HDR content is mastered.

DisplayCAL - Display Calibration and Characterization powered by ArgyllCMS
Current stable version 3.8.3 released 2019-07-02 | Changelog
DisplayCAL on Facebook
fhoech is offline  
post #693 of 6092 Old 04-25-2018, 05:08 AM
AVS Forum Special Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 1,047
Mentioned: 72 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 378
Sorry for the late reply.

Quote:
Originally Posted by madshi View Post
Yes, for negative XYZ I suppose simple clipping should do, although it should also be possible to desaturate just enough to get of the negative values.
Yes, I do desaturate.

Quote:
Originally Posted by madshi View Post
But do you have any special code for positive overflows (> 1.0)? If not, do you clip to 1.0? Or do you let ArgyllCMS handle the overflow?
It's taken care of before ArgyllCMS gets the data. If any component reaches > 1.0 I (have to) clip (which is handled by the desaturation mostly, as clipping in XYZ directly is a really bad idea), due to the ICCv2 encoding constraints I mentioned (where white is always at Y = 1) - you can ignore that particular aspect.

DisplayCAL - Display Calibration and Characterization powered by ArgyllCMS
Current stable version 3.8.3 released 2019-07-02 | Changelog
DisplayCAL on Facebook
fhoech is offline  
Sponsored Links
Advertisement
 
post #694 of 6092 Old 04-25-2018, 07:54 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by fhoech View Post
I fail to see why.
Because you keep ignoring the fact that 1) scientific doesn't clip if I raise MadVR's target nits to 600nits and 2) that dumb doesn't clip at say 500nits when scientific does.

This means that it's not the display clipping, but the highlights due to whatever scientific does in its tonemapping that dumb doesn't or does differently. Maybe scientific tries to fit more levels while dumb doesn't? I don't know, I can only report what I see.

I'll check that it's not my 3D LUT though just to rule it out. Because scientific is much redder than dumb, the clipping might come from the fact that that part of the color volume is more limited on my display rather than from the algo itself, i.e. orange/yellow will not clip on my display but red might. I don't think it's the case because it doesn't look like luminance or saturation clipping at the edge of the gamut, but more like levels clipping due to highlights compression.

Quote:
Originally Posted by fhoech View Post
After the BT2390 roll-off and tonemapping is applied, and the destination gamut is 2020 in madVR, the tonemapped values may well exceed the (roughly) effective P3 gamut of the projector (unless the tonemapping already makes sure that the tonemapped values can never exceed P3, which would seem odd).
Yes, I fully understand this, this is why I reported it, so that Madshi can investigate if scientific is doing something wrong compared to dumb in the examples I listed.

Quote:
Originally Posted by fhoech View Post
Well, in passthrough mode the original encoded values won't exceed P3 (on the signal side) because that's usually how HDR content is mastered.
Yes, but again it means that the display is able to display the content as it was mastered without clipping. If I use an HDR to SDR conversion, I want it to be as good or better as passthrough, because I'm overall fairly happy with passthrough, given that I can switch automatically between 2-3 custom curves according to the content, which MadVR doesn't do yet.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #695 of 6092 Old 04-25-2018, 08:22 AM
AVS Forum Special Member
 
fhoech's Avatar
 
Join Date: Apr 2013
Location: Stuttgart, Germany
Posts: 1,047
Mentioned: 72 Post(s)
Tagged: 0 Thread(s)
Quoted: 572 Post(s)
Liked: 378
Quote:
Originally Posted by Manni01 View Post
Because you keep ignoring the fact that 1) scientific doesn't clip if I raise MadVR's target nits to 600nits and 2) that dumb doesn't clip at say 500nits when scientific does.
Sure, I'm just saying that this is not a conclusive test as to where in the chain the clipping occurs (i.e. if scientific looks like it clips it might just be because it sends more saturated values which would still show detail if the display had an actual 2020 gamut). For a real conclusive test the display shouldn't do any extraneous processing on the signal (not even P3-within-2020 to actual display mapping, which it has to do if it receives 2020 data).
I just found it odd that there is any visible clipping at all in your case (if I set a 2020 target in madVR with the pixel shader parameters you're using, I'm not really seeing that, but then again, I can only look at the clips that were shared, not at any scenes outside of that). Maybe 'clipping' isn't the right word and you're referring to a certain 'flatness' that may be in some of the rolled-off areas, I'm not sure.

Quote:
Originally Posted by Manni01 View Post
I'll check that it's not my 3D LUT though just to rule it out. Because scientific is much redder than dumb, the clipping might come from the fact that that part of the color volume is more limited on my display rather than from the algo itself, i.e. orange/yellow will not clip on my display but red might. I don't think it's the case because it doesn't look like luminance or saturation clipping at the edge of the gamut, but more like levels clipping due to highlights compression.
Ok, fair enough.

DisplayCAL - Display Calibration and Characterization powered by ArgyllCMS
Current stable version 3.8.3 released 2019-07-02 | Changelog
DisplayCAL on Facebook
fhoech is offline  
post #696 of 6092 Old 04-25-2018, 08:44 AM
AVS Forum Addicted Member
 
Kris Deering's Avatar
 
Join Date: Oct 2002
Location: The Pacific Northwet
Posts: 10,608
Mentioned: 186 Post(s)
Tagged: 0 Thread(s)
Quoted: 3755 Post(s)
Liked: 6405
Quote:
Originally Posted by Manni01 View Post
The baseline is a freshly autocaled SDR BT2020 Gamma 2.4 calibration. I have a Calman Ultimate generated SDR BT2020 3D LUT 2.4 gamma (using my Discus profiled to my i1pro2), which brings the display to reference.

I set MadVR already to calibrated to BT2020 with a 2.4 gamma, and I enable the 3D LUT (or not) with HDR content converted to SDR.



The clipping in the highlights is to be expected as you lower the target nits and the highlights get compressed further, I'm only reporting that dumb is clipping far less than complex at the same target nits as you go below 600nits, which is an advantage to reproduce low APL shot as MadVR is not yet fully adaptive. You still have to choose a compromise between resolving the highlights in 4000nits titles (higher target nits) and reproducing low APL scenes properly in 1000nits or below titles (lower target nits). This needs to be addressed, but at a later stage, when we've settled on color. Complex Gamut Roll clips a little less, but the compromise in brightness/contrast isn't worth it compared to dumb or gamut clip.


You mention a 2020 3D LUT calibration with Calman. I’ve tried this numerous times with Calman (though not with the latest version yet) and I always end up with horrible results that track worse than just leaving 2020 alone after an Autocal with the JVC software. Can you provide your method and results for this? I’d be interested in knowing if your approaching the LUT differently. You can share via email or PM if you don’t want to disrupt the thread.


Sent from my iPhone using Tapatalk Pro

My Home Theater UPDATED DEC 2017
Technical Editor/Writer Sound and Vision Magazine
Deep Dive AV - Calibration, Consulting and Education
Kris Deering is offline  
post #697 of 6092 Old 04-25-2018, 10:01 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by fhoech View Post
Sure, I'm just saying that this is not a conclusive test as to where in the chain the clipping occurs (i.e. if scientific looks like it clips it might just be because it sends more saturated values which would still show detail if the display had an actual 2020 gamut). For a real conclusive test the display shouldn't do any extraneous processing on the signal (not even P3-within-2020 to actual display mapping, which it has to do if it receives 2020 data).
Yes, this is why I said I'll check with the 3D LUT off in order to rule out the 3D LUT itself. I'll also check if it happens with a DCI-P3 calibration, because in that case MadVR should know that the display cannot reach beyond P3 so shouldn't send anything beyond P3.

Quote:
Originally Posted by fhoech View Post
I just found it odd that there is any visible clipping at all in your case (if I set a 2020 target in madVR with the pixel shader parameters you're using, I'm not really seeing that, but then again, I can only look at the clips that were shared, not at any scenes outside of that). Maybe 'clipping' isn't the right word and you're referring to a certain 'flatness' that may be in some of the rolled-off areas, I'm not sure.
No, the flatness is with gamut roll due to the rolled-off areas, I'm talking about clipping detail in the highlights, as in detail goes away when you switch between dumb and scientific.

Quote:
Originally Posted by Kris Deering View Post
You mention a 2020 3D LUT calibration with Calman. I’ve tried this numerous times with Calman (though not with the latest version yet) and I always end up with horrible results that track worse than just leaving 2020 alone after an Autocal with the JVC software. Can you provide your method and results for this? I’d be interested in knowing if your approaching the LUT differently. You can share via email or PM if you don’t want to disrupt the thread.
Initially I did have some poor results, with lots of posterization, so I had given up, but I retried recently (I think it was with test build 5 and the latest Calman) when I recalibrated my PJ, and suddenly it was working as expected. I suspect MadVR more than Calman because the results were always correct when I checked the SDR BT2020 calibration in Calman, but I have no certitude. It could also be Calman, or it could be me. But now it works fine. That's one of the main advantages of MadVR's pixel shader vs passthrough, as now I can get a reference picture with HDR content too.

I simply profile my display with a BT2020 baseline in the JVC (BT2020 color profile, gamma 2.4) after an autocal, and I create an SDR BT2020 3D LUT in Calman with the same targets (BT2020, power gamma 2.4).

Of course I indicate in MadVR the 3D LUT to use for BT2020 content.

I use MadVR as a source and as a display though, so it might help. There might be a bug in Calman when you use the Radiance Pro as a source/display.

Anyway the result is near reference when checked in Calman (I only use a 1000points LUT). In MadVR the HDR picture is more accurate, especially gamma-wise. But it's very watchable without the 3D LUT, provided you've autocaled the SDR BT2020 profile.

Feel free to email me if you have any further questions, so as not to derail the thread.
Kris Deering likes this.

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

Last edited by Manni01; 04-25-2018 at 02:56 PM.
Manni01 is offline  
post #698 of 6092 Old 04-25-2018, 01:51 PM
Newbie
 
Join Date: Apr 2018
Posts: 1
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 1
With pictures...

Thank you very much for the work made in the latest version of MadVR on HDR->SDR conversion. I have spent some time comparing the options, and I thought I could help by sharing some results with screenshots illustrating the differences. Please excuse my english.

First, I found that "complex scientific mode - RGB clip" was the best algorithm, as it is the most faithful to original colors, what can be checked by increasing the target peak nits to see how they were intended, or by comparing with another player, like MPV (but the way it manages the conversion is far to be as good as MadVR). I noticed "complex scientific mode - gamut clip" has troubles with red areas, for example the burning wood.

Second, "dumb mode -color corrected" could be even better, as it is less prone to clipping on bright areas, if it did not make hightlights so white (it is visible on the flames).

Last, "30% luminance and 70% saturation" is the most balanced option, and "dynamic luminance & saturation reduction" often makes wrong choices, causing details to fade because gradient is lost (bright areas of a cloudy sky for example).

All test were made on a P2250 Syncmaster with a 300 target peak nits with 1000 nits movies. Hope it will help to improve algorithms or make right choices of remaining setting.
Attached Thumbnails
Click image for larger version

Name:	3 # RGB clip - 30&70.jpg
Views:	105
Size:	146.4 KB
ID:	2395500   Click image for larger version

Name:	3 # 1000 nits.jpg
Views:	93
Size:	107.9 KB
ID:	2395502   Click image for larger version

Name:	2 # RGB clip.jpg
Views:	105
Size:	150.4 KB
ID:	2395504   Click image for larger version

Name:	2 # gamut clip.jpg
Views:	99
Size:	149.7 KB
ID:	2395506   Click image for larger version

Name:	2 # convert gamut late.jpg
Views:	84
Size:	152.5 KB
ID:	2395508  

Click image for larger version

Name:	2 # color corrected.jpg
Views:	67
Size:	151.7 KB
ID:	2395510   Click image for larger version

Name:	3 # RGB clip - dynamic.jpg
Views:	62
Size:	146.2 KB
ID:	2395512  
mkohman likes this.
Antoniol is offline  
post #699 of 6092 Old 04-25-2018, 03:29 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by fhoech View Post
Sure, I'm just saying that this is not a conclusive test as to where in the chain the clipping occurs (i.e. if scientific looks like it clips it might just be because it sends more saturated values which would still show detail if the display had an actual 2020 gamut). For a real conclusive test the display shouldn't do any extraneous processing on the signal (not even P3-within-2020 to actual display mapping, which it has to do if it receives 2020 data).
I just found it odd that there is any visible clipping at all in your case (if I set a 2020 target in madVR with the pixel shader parameters you're using, I'm not really seeing that, but then again, I can only look at the clips that were shared, not at any scenes outside of that). Maybe 'clipping' isn't the right word and you're referring to a certain 'flatness' that may be in some of the rolled-off areas, I'm not sure.
I've run a few quick tests to rule out a few options.

First I have exactly the same behaviour when I disable the 3D LUT, so that's ruled out.

Second I tried a quick DCI-P3 calibration, using my DCI-P3 profile with a 2.4 gamma and telling MadVR the display was already calibrated to P3 2.4, I didn't have the time to run a 3D LUT). I had exactly the same results.

Looking closely at the picture, I think what happens at 500nits is that dumb has red but also yeallow/orange in the shot, while scientific replaces all the orange/yellow with red, at very similar saturation/luminance as the red already in the shot. This looks like clipping, but I think it's more a case of "same red" than clipping per se at the edge of the gamut. As I said earlier, it doesn't necessarily happen on the most saturated/brightest reds in the shot, it simply happens to be more red replacing the orange/yellow showing on dumb. So red on red = red everywhere, hence less detail than in dumb.

If I use 600nits target, there is more headroom for the highlights and scientific shows more levels of red, which means there is less clipping (if any).

Does it make sense?

Again, that shot *should* probably be shown with a 600nits target, as it's a 4,000nits content going up to 10,000nits. A brightness factor of 5 (diffuse white of 20) is what I would normally use for that type of content with my 120nits display.

But for some reasons, as has always been the case, dumb/naive is less prone to clipping, which allows to lower the target to 500nits while keeping enough details in the highlights, resulting if far better results in darker titles, such as the infamous 6nits shot in The Revenant.

I'll need to do more tests with other 4,000nits titles, to see if this "red on red clipping" also happens with other 4,000nits+ titles. The Shallows and Secret Life of Pets are not good examples because there are very few explosions/fires at 4,000nits+, given that they are 2,500/1,000nits titles.

The Dark Knight would have been great but MaxCll is barely above 1000nits, so none of these issues would show.

I tried with BvsS quickly and I have exactly the same issue. For example, look at 02:28:16, around the eyes, the red clips (detail is lost) because the orange is replaced by red and red on red = less detail. On that shot in BvsS, there is clipping (loss of detail) even at 600nits in scientific compared to dumb. At 500nits it gets worse and it adds ugly contouring. Dumb keeps a lot more detail (and no contouring) even at 500nits.

As soon as we get profiles allowing us to test for content max brightness, this won't be a problem anymore as we'll be able to swap profiles according to content.

In the meantime, as I don't fancy changing things manually for each title, I prefer to use a compromise that works well for all titles, and that's dumb convert gamut late with a target of 500nits. I actually enjoy the overall boost to the picture brightness with a brightness factor of 4 (diffuse white of 25) vs my usual 5 (diffuse white at 20).

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

Last edited by Manni01; 04-25-2018 at 03:34 PM.
Manni01 is offline  
post #700 of 6092 Old 04-25-2018, 03:46 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Regarding the above post, I've loaded for the first time the Masciola pattern to compare what scientific does vs dumb with a 500nits target, and I guess I have the answer.

Scientific clips everything in the middle of the pattern, on most channels, more or less badly.

On red, it clips everything between 500nits and 1500nits. On Green, between 500nits and 800nits. Blue doesn't clip much in the middle, and neither does yellow.

So maybe that's part of the explanation?

Dumb does clip above 2000nits or so, but there is no clipping in the middle, and certainly not as low as 500nits.

So if red clips as early as 500nits in scientific (due to this bumb in the middle) and only at 2000nits in dumb, that might explain some of what I'm saying (orange/yellow in dumb replaced by red that should be different but isn't).

We've already reported this bumb in scientific but I had forgotten about it.

If I switch target nits to 600, there is still such bump in the middle with scientific, but it's much reduced. Maybe 700-1000nits instead of 500-1500nits.

As Madshi had explained very well why the Masciola pattern doesn't necessarily show what happens with real content, I didn't want to look at patterns before looking at content because it's easy to get influenced that way.

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

Last edited by Manni01; 04-25-2018 at 03:54 PM.
Manni01 is offline  
post #701 of 6092 Old 04-25-2018, 06:00 PM
AVS Forum Special Member
 
stevenjw's Avatar
 
Join Date: Jul 2001
Location: Palm Coast, Florida, USA
Posts: 2,938
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 635 Post(s)
Liked: 829
Quote:
Originally Posted by Manni01 View Post
nVidia levels (as well as many other things) are broken post 385.28, at least in Windows 10 build 1709. They are inverted in 390.65 (you need to set MadVR to 0-255 to get 16-235 if your display is set to limited), and fully borked in 391.xx. Revert to 385.28 and check again. I'm using 385.28 and there is no issue with black levels.
Sorry to sidetrack this very important work and thread, but you're not accepting PMs. I just purchase and downgraded firmware for "friendly" LG drive and my GTX 1080 card showed up today. I plan to install tonight and have downloaded 385.28 driver. Can you PM or post what specific link(s) you're following concerning NVidia drivers and this issue? I'd like to follow them as well to know when it's "safe" to finally upgrade from this driver that's from last August. Thanks! And keep up the great work!

...Steve
"Opinions are like orgasms… mine matters most and I really don’t care if you have one or not." ;)
 
My HT gear
stevenjw is offline  
post #702 of 6092 Old 04-25-2018, 06:07 PM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Quote:
Originally Posted by stevenjw View Post
Sorry to sidetrack this very important work and thread, but you're not accepting PMs. I just purchase and downgraded firmware for "friendly" LG drive and my GTX 1080 card showed up today. I plan to install tonight and have downloaded 385.28 driver. Can you PM or post what specific link(s) you're following concerning NVidia drivers and this issue? I'd like to follow them as well to know when it's "safe" to finally upgrade from this driver that's from last August. Thanks! And keep up the great work!
Just set your Nvidia drivers to 16-235 12bit 3840/23p limited range and MadVR to full range 0-255.

I know its not the most recommended, but I have been doing it forever and literally never ever see banding issues.
stevenjw likes this.

JVC X9500 (RS620) | 120" 16:9 | Marantz AV7702 MkII | Emotiva XPA-7 | DIY Modular Towers | DIY TPL-150 Surrounds | DIY Atmos | DIY 18" Subs
-
MadVR Settings | UHD Waveform Analysis | Arve Tool Instructions + V3 Javs Curves
Javs is offline  
post #703 of 6092 Old 04-25-2018, 11:49 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by stevenjw View Post
Sorry to sidetrack this very important work and thread, but you're not accepting PMs. I just purchase and downgraded firmware for "friendly" LG drive and my GTX 1080 card showed up today. I plan to install tonight and have downloaded 385.28 driver. Can you PM or post what specific link(s) you're following concerning NVidia drivers and this issue? I'd like to follow them as well to know when it's "safe" to finally upgrade from this driver that's from last August. Thanks! And keep up the great work!
Best is to follow MadVR's thread on Doom9, but unless you're a gamer there is no real reason to update the driver on a HTPC.

Quote:
Originally Posted by Javs View Post
Just set your Nvidia drivers to 16-235 12bit 3840/23p limited range and MadVR to full range 0-255.

I know its not the most recommended, but I have been doing it forever and literally never ever see banding issues.
If you do this, you let nVidia's driver dither behind MadVR's back, so it's indeed not recommended, but as long as you have no banding issues, good for you.
stevenjw likes this.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #704 of 6092 Old 04-26-2018, 12:22 AM
AVS Forum Special Member
 
Javs's Avatar
 
Join Date: Dec 2012
Location: Sydney
Posts: 7,849
Mentioned: 475 Post(s)
Tagged: 0 Thread(s)
Quoted: 6791 Post(s)
Liked: 6418
Quote:
Originally Posted by Manni01 View Post
If you do this, you let nVidia's driver dither behind MadVR's back, so it's indeed not recommended, but as long as you have no banding issues, good for you.
Yep, I even tested it at length and the results using full range vs limited were identical.

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 #705 of 6092 Old 04-26-2018, 06:06 PM
Senior Member
 
bcec's Avatar
 
Join Date: Aug 2013
Location: Bay Area, CA
Posts: 315
Mentioned: 5 Post(s)
Tagged: 0 Thread(s)
Quoted: 238 Post(s)
Liked: 123
Quote:
Originally Posted by Javs View Post
Yep, I even tested it at length and the results using full range vs limited were identical.
I don't want to derail the conversation from HDR but here's another thing to consider:

16-235 to 0-255 is simply a mapping of 16-235 values into 0-255 ones. The result of this mathematical mapping is that you end up with floating point numbers but they will have to be rounded to integer values when sent through HDMI. For instance, 19 will mathematically correspond to 3.49, but will be sent as 3. 20 will correspond to 4.65, but will be sent as 5. You will never have 4, and what's more, due to rounding, you will get a gradient (3 instead of 3.49, 5 instead of 4.65 etc) that is not 100% accurate to the source.

So, the best setup for video (if you don't care about crushed blacks on desktop) is:
GPU outputs 0-255
Video player outputs 16-235
Display expects 16-235


Having said that, most people, including myself use:
GPU outputs 0-255
Video player outputs 0-255
Display expects 0-255
stevenjw, sonichart and Manni01 like this.
bcec is offline  
post #706 of 6092 Old 04-26-2018, 10:08 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by bcec View Post
So, the best setup for video (if you don't care about crushed blacks on desktop) is:
GPU outputs 0-255
Video player outputs 16-235
Display expects 16-235
Yep, this is why I'm using full / limited / limited. I'd rather have MadVR's carefully crafted levels untouched than have nVidia dither behind its back. I don't care about the desktop, my HTPC is for film playback first, and I don't care about games, so no downsides to use 385.28. Plus I use REW and Asio4All compatibility is broken from 390.x

Now let's go back to our normal programming
stevenjw, sonichart and bcec like this.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #707 of 6092 Old 04-27-2018, 01:01 AM
Member
 
((( atom )))'s Avatar
 
Join Date: Jan 2010
Posts: 75
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 38 Post(s)
Liked: 28
Very interesting discussion. I do have some banding, so I really might have been fooled by either the wrong levels behind NVIDIAS limited/full button or I might have caught a silent driver update which broke things. Still didn't find the time to check it out, but hopefully on the weekend.

So the only thing broken (for video only usage) is the wrong labels on the level switch? Meaning, I could simply use limited instead of full to get correct full range output?
((( atom ))) is online now  
post #708 of 6092 Old 04-27-2018, 01:31 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by ((( atom ))) View Post
Very interesting discussion. I do have some banding, so I really might have been fooled by either the wrong levels behind NVIDIAS limited/full button or I might have caught a silent driver update which broke things. Still didn't find the time to check it out, but hopefully on the weekend.

So the only thing broken (for video only usage) is the wrong labels on the level switch? Meaning, I could simply use limited instead of full to get correct full range output?
Please let’s go back to topic or this thread is going to turn into a general MadVR support thread. Madshi doesn’t need this. All this has been discusssed in MadVR’s thread on Doom9, so that’s the place to read and ask such generic questions. I posted in that thread recently a list of what was broken (for me) in various nVidia drivers after 385.28. Thanks

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #709 of 6092 Old 04-27-2018, 04:52 AM
Advanced Member
 
hasta666's Avatar
 
Join Date: Apr 2016
Posts: 622
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 461 Post(s)
Liked: 152
I watched The Death Cure yesterday and the colors looked so wrong that I had to tweak settings for 20 minutes before going back to the movie itself.

I’m using a BT.2020 3D LUT. I was using the scientific mode in the beginning. Everything looked hella red, skins not close to natural. Using dumb mode late gamut was the best looking picture, most natural. Not saying that it was the closest to what was intended by the colorist though. I need to watch the SDR Blu-ray and do a comparison. I also tried HDR passthrough and in terms of colors, it was pretty good and looked natural. Unfortunately in passthrough the picture is too dark plus I can’t really use it for every day watch since I have to enable HDR in Windows.

In the end I’m most please with scientific mode but with some movies it just don’t look right. Dumb mode is not better, except for late gamut which best every other mode. What is strange though is that I only got that problem for certain titles. In the end I’m just wondering if those titles are supposed to look like sh*t or if something’s wrong in my chain. Still it’s pretty annoying to buy a title for such result.
hasta666 is offline  
post #710 of 6092 Old 04-27-2018, 05:47 AM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by hasta666 View Post
I watched The Death Cure yesterday and the colors looked so wrong that I had to tweak settings for 20 minutes before going back to the movie itself.

I’m using a BT.2020 3D LUT. I was using the scientific mode in the beginning. Everything looked hella red, skins not close to natural. Using dumb mode late gamut was the best looking picture, most natural. Not saying that it was the closest to what was intended by the colorist though. I need to watch the SDR Blu-ray and do a comparison. I also tried HDR passthrough and in terms of colors, it was pretty good and looked natural. Unfortunately in passthrough the picture is too dark plus I can’t really use it for every day watch since I have to enable HDR in Windows.

In the end I’m most please with scientific mode but with some movies it just don’t look right. Dumb mode is not better, except for late gamut which best every other mode. What is strange though is that I only got that problem for certain titles. In the end I’m just wondering if those titles are supposed to look like sh*t or if something’s wrong in my chain. Still it’s pretty annoying to buy a title for such result.
Did you try to disable your 3D LUT? There shouldn’t be a huge difference between “this display is already calibrated with BT2020 gamma 2.4” and passthrough, provided you use the same parameters.

Passthrough is only dark because you have not designed a good custom curve for your set-up. There is no gain with MadVR re dark titles (not yet), you’re simply using a target nits in pixel shader that means a lower brightness factor (higher diffuse white) than whatever you are using in passthrough. Not the place to discuss this though.
sonichart likes this.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #711 of 6092 Old 05-04-2018, 06:26 PM
Advanced Member
 
mkohman's Avatar
 
Join Date: Jun 2009
Location: London, United Kingdom
Posts: 769
Mentioned: 20 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 340
Guys, I have been following this thread for a little while but I need some help please. I want to try out this HDR to SDR with madVR but not sure how or what I need to do? I have a JVC x5900 and the HDR is ok but I think this may be better from what I have been reading. Apologies if this has already been covered but is there a quick guide on what settings I need to apply within madVR to get this working ? Also do I need to add like setting or a file etc? Thank you so much in advance for all your help!
mkohman is offline  
post #712 of 6092 Old 05-05-2018, 12:04 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 609
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 330 Post(s)
Liked: 299
Quote:
Originally Posted by mkohman View Post
Guys, I have been following this thread for a little while but I need some help please. I want to try out this HDR to SDR with madVR but not sure how or what I need to do? I have a JVC x5900 and the HDR is ok but I think this may be better from what I have been reading. Apologies if this has already been covered but is there a quick guide on what settings I need to apply within madVR to get this working ? Also do I need to add like setting or a file etc? Thank you so much in advance for all your help!
As has been repeatedly stated, this is not the appropriate thread for that question. There are others already set up to answer your question and provide guidance. Please use those threads. For example: https://forum.doom9.org/showthread.php?t=146228
stevenjw likes this.
ddgdl is offline  
post #713 of 6092 Old 05-06-2018, 01:08 AM
Advanced Member
 
mkohman's Avatar
 
Join Date: Jun 2009
Location: London, United Kingdom
Posts: 769
Mentioned: 20 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 340
Quote:
Originally Posted by ddgdl View Post
As has been repeatedly stated, this is not the appropriate thread for that question. There are others already set up to answer your question and provide guidance. Please use those threads. For example: https://forum.doom9.org/showthread.php?t=146228
Thank you I will check out that thread now

Sent from my Moto G (4) using Tapatalk
mkohman is offline  
post #714 of 6092 Old 05-06-2018, 01:42 AM
Advanced Member
 
mkohman's Avatar
 
Join Date: Jun 2009
Location: London, United Kingdom
Posts: 769
Mentioned: 20 Post(s)
Tagged: 0 Thread(s)
Quoted: 398 Post(s)
Liked: 340
Quote:
Originally Posted by ddgdl View Post
As has been repeatedly stated, this is not the appropriate thread for that question. There are others already set up to answer your question and provide guidance. Please use those threads. For example: https://forum.doom9.org/showthread.php?t=146228
I can't seem to register to doom9 in order to post my question. I would really appreciate if anyone could please point me to the right direction in how to apply this HDR to SDR improvement thing on madVR . Thank you so much in advance.. Is there a Thread on AVS where I can get help? Thanks
mkohman is offline  
post #715 of 6092 Old 05-06-2018, 04:02 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 609
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 330 Post(s)
Liked: 299
Quote:
Originally Posted by mkohman View Post
Quote:
Originally Posted by ddgdl View Post
As has been repeatedly stated, this is not the appropriate thread for that question. There are others already set up to answer your question and provide guidance. Please use those threads. For example: https://forum.doom9.org/showthread.php?t=146228
I can't seem to register to doom9 in order to post my question. I would really appreciate if anyone could please point me to the right direction in how to apply this HDR to SDR improvement thing on madVR . Thank you so much in advance.. Is there a Thread on AVS where I can get help? Thanks
Here: https://www.avsforum.com/forum/26-hom...rt-thread.html
mkohman likes this.
ddgdl is offline  
post #716 of 6092 Old 05-10-2018, 11:38 AM
Advanced Member
 
ddgdl's Avatar
 
Join Date: Mar 2007
Posts: 609
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 330 Post(s)
Liked: 299
@madshi have you seen the results Jim has gotten with a desaturation algo in the radiance?
@KarlKlammer has posted some screenshots today over in the radiance thread: https://www.avsforum.com/forum/showthread.php?t=2172017
Soulnight likes this.
ddgdl is offline  
post #717 of 6092 Old 05-16-2018, 03:53 AM
Member
 
stefanelli73's Avatar
 
Join Date: Jan 2008
Posts: 86
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 21 Post(s)
Liked: 15
I own an OPTOMA UHZ 65 laser video projector that handles very well the HDR, in fact with madvr I'm going to passthrough .... yesterday out of curiosity I wanted to use the HDR to SDR conversion, the image is more brighter, but the colors appear washed out, even if you set it in "this display is already calibrated" BT.2020, what can it depend on?
stefanelli73 is offline  
post #718 of 6092 Old 06-03-2018, 04:20 PM
AVS Forum Special Member
 
stevenjw's Avatar
 
Join Date: Jul 2001
Location: Palm Coast, Florida, USA
Posts: 2,938
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 635 Post(s)
Liked: 829
Haven't seen any progress since late April. I'm hoping that this effort is not dead. Currently still using custom curves and MadVR HDR Passthru, but hoping that will once I read that the work is done or nearly so with MadVR TM for projectors.

...Steve
"Opinions are like orgasms… mine matters most and I really don’t care if you have one or not." ;)
 
My HT gear
stevenjw is offline  
post #719 of 6092 Old 06-03-2018, 05:30 PM
AVS Forum Special Member
 
Manni01's Avatar
 
Join Date: Sep 2008
Posts: 9,014
Mentioned: 310 Post(s)
Tagged: 0 Thread(s)
Quoted: 5361 Post(s)
Liked: 5421
Quote:
Originally Posted by stevenjw View Post
Haven't seen any progress since late April. I'm hoping that this effort is not dead. Currently still using custom curves and MadVR HDR Passthru, but hoping that will once I read that the work is done or nearly so with MadVR TM for projectors.
This effort is not dead, Madshi just has a lot of commercial work on his desk at the moment, so little time for MadVR. He will come back to this as soon as he has more time to dedicate to it.
stevenjw and neo_2009 like this.

JVC Autocal Software V11 Calibration for 2019 Models (Google)
Batch Utility V4.02 May 16 2019 to automate measurements files for madVR with support for BD Folders
Manni01 is offline  
post #720 of 6092 Old 06-04-2018, 10:12 AM
AVS Forum Special Member
 
stevenjw's Avatar
 
Join Date: Jul 2001
Location: Palm Coast, Florida, USA
Posts: 2,938
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 635 Post(s)
Liked: 829
Thanks! It's good to know that he's still on the case. I guess even Superman had to take the cape off once in a while and pay the bills as Clark Kent.
Willie and Manni01 like this.

...Steve
"Opinions are like orgasms… mine matters most and I really don’t care if you have one or not." ;)
 
My HT gear
stevenjw 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