madVR - high quality video renderer (GPU assisted) [Archive] - Page 732

FDisk80

28th February 2016, 12:06

Oh yes, there were some users who had problems with presentation queues not properly filling, e.g. when using 10bit output. I think this new option *may* work around that. But if it does, I wonder if it affects audio/video sync?

That was me. :) That's why I got interested in this option. Will test it out. :cool:
Actually it didn't matter if it was 10bit or 8bit, the FSE mode itself gave me strange queues problems.

omarank

28th February 2016, 14:40

* added "use alternative glitch handling mode" option for D3D11 presentation

Is it possible that 10 bit frame packed 3D can work with this alternative mode or MS APIs don't allow 10 bit for 3D at all?

aufkrawall

28th February 2016, 15:04

Regarding missing feedback for sharpen filters:
I must admit that my interest in sharpen filters for videos has quite declined, thanks to SuperRes. I have to upscale most videos to WQHD, so I almost always have it as a post scaling option.
My problem also is that applying sharpen as image enhancement and then doing SuperRes after upscaling looks very strange to my eyes, the artificial look that is introduced by sharpen gets boosted.
When I'm watching bad/soft sources (ARD Mediathek, wtf :eek: ), I was already very happy with the old sharpeners applied after SuperRes, sharpen edges or adaptive sharpen. I quite started to like adaptive sharpen again after bacondither's improvements regarding brightness etc.

Well, maybe I will find some images that seem well suited for testing to me...

ang3l

28th February 2016, 15:16

* fixed: SSIM AR eventually left some resources open

just to confirm that also for me is fixed, thx

markanini

28th February 2016, 15:29

SuperRes: 3 + Enhance details: 1.0 is amazing for resolution but I get trouble with ringing. I found a compromise in using NEDI for image doubling but I doesn't feel ideal.

aufkrawall

28th February 2016, 15:31

NEDI? Best for ringing reduction is NNEDI3.

madshi

28th February 2016, 15:31

Forgot the default wasn't with the key, same file I've linked you a few times, here (https://mega.nz/#!rI5BRRqD!49WQf6LMMEUQpyEz-Ug8aL-OE6IkB7ZRNOagD8WOhnQ).
Ok, thanks, I can reproduce some problems, at least, will have a look at that.

That was me. :) That's why I got interested in this option. Will test it out. :cool:
Actually it didn't matter if it was 10bit or 8bit, the FSE mode itself gave me strange queues problems.
Ok, let me know how the new option works for you! Please also check whether audio/video sync is and stays in sync. It should, but it's worth a test.

Is it possible that 10 bit frame packed 3D can work with this alternative mode or MS APIs don't allow 10 bit for 3D at all?
Possibly. I'll have to check, currently I force 8bit for 3D.

Regarding missing feedback for sharpen filters:
I must admit that my interest in sharpen filters for videos has quite declined, thanks to SuperRes. I have to upscale most videos to WQHD, so I almost always have it as a post scaling option.
My problem also is that applying sharpen as image enhancement and then doing SuperRes after upscaling looks very strange to my eyes, the artificial look that is introduced by sharpen gets boosted.
When I'm watching bad/soft sources (ARD Mediathek, wtf :eek: ), I was already very happy with the old sharpeners applied after SuperRes, sharpen edges or adaptive sharpen. I quite started to like adaptive sharpen again after bacondither's improvements regarding brightness etc.

Well, maybe I will find some images that seem well suited for testing to me...
I remember that you had a strong dislike against running FineSharp (crispen edges) with linear light turned on. Would be great if you could test whether crispen edges with LL gets your blessing now that the new sharpening AR filter is available. Also "sharpen images" has a bit different look now, because the AR filter method has changed (it lets more sharpening through).

madshi

28th February 2016, 15:32

SuperRes: 3 + Enhance details: 1.0 is amazing for resolution but I get trouble with ringing. I found a compromise in using NEDI for image doubling but I doesn't feel ideal.
Which is causing the ringing? SuperRes or Enhance details? Or the upscaling algorithm?

markanini

28th February 2016, 15:48

Which is causing the ringing? SuperRes or Enhance details? Or the upscaling algorithm?

I'm gonna take back that anything is causing ringing, but rather making source ringing more obvious. SuperRes and Enhance details both contribute to this effect. NEDI is soft but retains resolution, obviously not as good as NNEDI3 at joining lines. I find using NEDI looks better than turning with down SuperRes with NNEDI3.

leeperry

28th February 2016, 15:48

NEDI? Best for ringing reduction is NNEDI3.
NNEDI3 is way too sharp IME in combination with SR+SSIM, NEDI provides a nice in-between NNEDI3 & plain Jinc upscaling.

SuperRes: 3 + Enhance details: 1.0 is amazing for resolution but I get trouble with ringing. I found a compromise in using NEDI for image doubling but I doesn't feel ideal.
There's no free-lunch to be expected IME, as much as mVR does miracles it can't magically turn SD into HD.

"SSIM 1D" is the same as "SSIM" in v0.90.9 (just a bit faster, I hope). It's based on a Bicubic kernel. "SSIM 2D" is based on a Jinc kernel, so it's rather slow.
Oh that figures why it's sharper and more natural looking then and it might not be such a big of a deal because I seem to really only find "SSIM 2D 25%" useful on doubled SD@1080p as it would appear too sharp for 720p@1080p even in 1D at 25%.

markanini

28th February 2016, 16:02

It's hard to tell the difference between crispen edges in LL or GL. I prefer the less resource heavy option.
Regarding the more aggressive downsampling AR I opened a few hi-res stills and from memory all I can say is that it doesn't look any worse than before. I'm using SSIM since it was introduced.

There's no free-lunch to be expected IME, as much as mVR does miracles it can't magically turn SD into HD.

I knew I was pushing MadVR to it's limits ;) Still I wonder if NNEDI3 or the other algos could be improved.

Georgel

28th February 2016, 16:53

Here's a better way to get to what you have done.

Download Nvidia Inspector:
http://www.techspot.com/downloads/5077-nvidia-inspector.html
Click the Profile Settings button:
http://i.imgur.com/gLvT5wB.png
In the Profiles box, select Media Player Classic
Scroll down to "Common" (Blue text with a horizontal line)
Under Common you will see "Power management mode"
To the right of that you will see "Adaptive"
Click in that box and select "Prefer maximum performance"
Scroll down to "Other" (Blue text with a horizontal line)
Under Other you will see "Enable application for Optimus"
To the right of that you will see a bunch of capital letters
Click in that box and select SHIM_RENDERING_MODE_ENABLE
Click Apply Changes

It depends on what you consider better, and on hardware setup.

I used Nvidia Ispector method for a while, but the settings are not saved on certain drivers, and this caused me to need to go to Nvidia Inspecor every time I restarted laptop.

I think that this was dependent upon the certain combination of drivers and settings that I have, but the method I discovered later would work on all hardware combinations I tested it.

aufkrawall

28th February 2016, 17:23

NNEDI3 is way too sharp IME in combination with SR+SSIM, NEDI provides a nice in-between NNEDI3 & plain Jinc upscaling.

When upscaling low-res content to WQHD, everything except of NNEDI3 quickly turns out to be simply rubbish in comparison, at least for the initial image doubling step.
NEDI is by far more aliased than NNEDI3, even though it is much softer.
So, what's the point of having more artifacts and a softer image at the same time? Doesn't make sense to me.
Why would NNEDI3 be "too sharp"?

omarank

28th February 2016, 17:46

Possibly. I'll have to check, currently I force 8bit for 3D.
Ok. I hope it works.

chros

28th February 2016, 19:46

It depends on what you consider better, and on hardware setup.
....
I think that this was dependent upon the certain combination of drivers and settings that I have, but the method I discovered later would work on all hardware combinations I tested it.
+1
And I don't like this, especially with laptops: "Click in that box and select "Prefer maximum performance"". I'm using "Adaptive" and I restrict the GPU usage to achieve less heat/noise with nvidiainspecor via command line at logon (through Task Scheduler):
"C:\Program Files (x86)\nVidiaInspector\nvidiaInspector.exe" -setGpuClock:0,1,470 -setMemoryClock:0,1,800 -setPstateLimit:0,5

leeperry

28th February 2016, 19:51

I knew I was pushing MadVR to it's limits ;) Still I wonder if NNEDI3 or the other algos could be improved.
IIRC NNEDI3's magic is essentially a black box whose secrets are only known by tritical with coeffs cooked using a whole bunch of reference material.

When upscaling low-res content to WQHD, everything except of NNEDI3 quickly turns out to be simply rubbish in comparison, at least for the initial image doubling step.
NEDI is by far more aliased than NNEDI3, even though it is much softer.
So, what's the point of having more artifacts and a softer image at the same time? Doesn't make sense to me.
Why would NNEDI3 be "too sharp"?
I would guess that the higher the output resolution the more NNEDI3 becomes a pre but IIRC at some point even madshi said that sxbr+SR was a good alternative to it. SR works its magic and so does SSIM 2D afterwards, surely things would change w/o SR.

I've always found NNEDI3 too digital looking and I really don't like the oil painting look it can give sometimes. Sharp but no free-lunch either, anyway at this point all we gotta do is pick our poison. I'm certainly very impressed by what I see myself :)

markanini

28th February 2016, 20:32

I'm sticking to NEDI with SR+Enhance details for the time being cause I hate Gibbs artifacts more than anything else. Maybe for clean sources I'll swtich to NNEDI3 temporarily.

adhara

28th February 2016, 20:43

@Warner306

In your guide around madvr, you say for example about supersampling (or image doubling):

1080p -> 2160p Supersampling (for high-end GPUs):
Chroma: super-xbr100 + AR
Image: Jinc3 + AR
Double Luma: 2x or greater - super-xbr100
Double Chroma: 2x or greater - super-xbr100
Downscaling: SSIM 1D 100% + AR + LL
Upscaling refinement: crispen edges (1.0) + AR, SuperRes (3)
Artifact removal - Debanding: Medium/High
Image enhancements: Off
Dithering: Ordered

I noted (with CTRL-J during playback) that the image doubling/quadrupling processing is applied only when "always - if upscaling is needed" option is checked.
If I only select "only if scaling factor..." option, doubling and/or quadrupling, the processing is not applied.

How come ?

aufkrawall

28th February 2016, 20:56

SR works its magic and so does SSIM 2D afterwards, surely things would change w/o SR.

I haven't really noticed a difference yet whether SSIM is used for downscaling or bicubic 150 after image doubling. However, I don't use image doubling for factors <1.5 since I never saw any benefit by it over Jinc AR + SR.
<1.5 for me e.g. is 1080p -> WQHD, so usually good sources.

I've always found NNEDI3 too digital looking and I really don't like the oil painting look it can give sometimes.

I don't like this look anymore either, but have you tried SuperRes against it? I think it works wonders, and unlike with super-xbr, haloing, ringing, aliasing and dark line boosting doesn't become an issue.
I tried various photographs and NNEDI3 + SuperRes always came closest to ground truth for me. And btw. even with a SR strength of 4, result was still much softer than ground truth.

MS-DOS

28th February 2016, 21:12

madshi,

Could you make madVR not render frames, or render at extremely low framerates (1 fps), or at 1x1 resolution when the player is minimized? I often listen to various videos\streams while playing games at fullscreen, periodically alt-tabbing when something interesting is happening, but even when mpc-hc is minimized it still consumes quite a lot of gpu power with madVR, so I have to switch to VMR-9\EVR :(

macycat

28th February 2016, 21:17

There have been some comments on performance issues including dropped frames, presentation glitches, etc., that seem to be related to the power state of the video card as per this recent sub-thread:

+1
And I don't like this, especially with laptops: "Click in that box and select "Prefer maximum performance"". I'm using "Adaptive" and I restrict the GPU usage to achieve less heat/noise with nvidiainspecor via command line at logon (through Task Scheduler):
"C:\Program Files (x86)\nVidiaInspector\nvidiaInspector.exe" -setGpuClock:0,1,470 -setMemoryClock:0,1,800 -setPstateLimit:0,5

I recently noticed problems when I tried to use the Direct3D 11 presentation path. I eventually selected "Prefer maximum performance" in the general settings for my nVidia GTX670 video card. Since in my HTPC application I don't use a laptop, my rig is always powered. Therefore, it is not a big deal for me to just select "Prefer maximum performance" and not use "Adaptive". Based on this sub-thread, I selected "Prefer maximum performance" only for both the 32-bit and 64-bit versions of MPC-HC application and selected "Adaptive" for the global settings. However, I really didn't notice any difference in fan noise when I had "Prefer maximum performance" selected in the global settings versus "Adaptive".

Although something like this was discussed back November, a good reminder for those those having weird frame drops and/or presentation glitches when using a decent video card and not maxing out the madVR settings, make sure that the video card isn't in a low power state.

aufkrawall

28th February 2016, 21:23

FDisk80

28th February 2016, 21:34

Ok, let me know how the new option works for you! Please also check whether audio/video sync is and stays in sync. It should, but it's worth a test.

Tested. Don't see any difference at all. Present Queue drops the same with or without this option enabled when the seek bar is on screen in FSE mode. Also same stutter as before when seek bar is visible.

Didn't notice any sync problems.

Warner306

28th February 2016, 21:57

@Warner306

In your guide around madvr, you say for example about supersampling (or image doubling):

I noted (with CTRL-J during playback) that the image doubling/quadrupling processing is applied only when "always - if upscaling is needed" option is checked.
If I only select "only if scaling factor..." option, doubling and/or quadrupling, the processing is not applied.

How come ?

That's a copy and paste error. You have to set it to "always."

XMonarchY

28th February 2016, 22:49

Sorry to go off-topic about H.264 and HEVC, but why would it not become a standard even for HD content? It let's you make lower-sized files at the same quality as H.264 OR let's you make same-size files, but at higher quality than H.264. Is it a licensing problem? I don't know how licensing works. Does ripping software pay for it or what? I know I paid for Bandicam that support HEVC on-the-fly encoding during videogame playback. Bandicam isn't that expensive...

Stereodude

28th February 2016, 22:53

There are a couple of IMAX movies, which switch between Cinemascope and 16:9, but those are not too common.
I think it's a little more common with 3D movies. For example, Guardians of the Galaxy changes aspect ratio throughout the movie, but only the 3D version.

QBhd

28th February 2016, 23:29

It's been discussed by a few of us here, but no one seems to be able to track this down. Since v0.90.11 playback results in a black screen for me. I can pull up PotPlayer's OSD and madVR's OSD... sound plays and so does the video, except that the video is just black. Here is the screen shots which clearly shows something is broken. First shot is of v0.90.10 and the second is v0.90.13 (11 & 12 were the same)

http://i.imgur.com/fumZWr1.jpg

http://i.imgur.com/H5fYYXb.jpg

I don't want this issue to leave me out of any new improvements...

QB

Sparktank

29th February 2016, 00:03

I think it's a little more common with 3D movies. For example, Guardians of the Galaxy changes aspect ratio throughout the movie, but only the 3D version.

It's only becoming common with recent movies in 3D.

Christopher Nolan only works in 2D.
The Dark Knight
The Dark Knight Rises
Interstellar

Star Trek: Into Darkness was re-released with alternating aspect ratios in 2D only, despite being a 3D movie, in The Compendium bundle.

Tron: Legacy has alternating AR in 2D as well as 3D.

For 3D-only with alternating AR, Transformers: Age of Extinction.
Yet, Revenge of the Fallen was a Wal-Mart exclusive in 2D-only for alternating AR.

The 3D-only with alternating AR is only a recent convention. I believe it's the director's new decision. "How it was meant to be seen", as they were filmed for IMAX and 3D (regardless of post-production conversion for 3D).
Home entertainment is just a bonus. Movies are meant for cinema only.

But this is off topic now.
------------------------

RE:
Image Enhancements: crispen edges + LL
I don't see much of a difference on or off, even with a level of 4.0.
Anti-ringing definitely helps with the crispen edges, regardless of LL.

macycat

29th February 2016, 01:33

It should not be necessary to select maximum performance profile, since the load of madVR is very constant, compared to games. So there really shouldn't be any confusion by the driver to pick the appropriate clock state.
I suspect if there is a problem, it might be related to a driver bug.

I have heard that before, so I did a quick test with some 1080i content that I typically watch. With a custom application setting of ""Prefer maximum performance", my rendering time is ~6 ms. When I change the setting to "Adaptive", the rending time goes up to 13 ms. Something is impacting the rendering times. My typical settings include ordered diffusion, Direct3D 11, Exclusive mode, and display at 1080i (video card is doing the scaling).

My display is natively 2560x1440p60. I played around with letting madVR do the scaling. I did the same test and obtained ~13 ms rendering times with either"Prefer maximum performance" or "Adaptive" power mode settings.

:confused:

jmelan

29th February 2016, 01:52

I don't do anything specific to support NVidia 3D Vision specific hardware. All I'm doing is using the OS 3D APIs to do 3D playback. Whether that works or doesn't work with NVidia 3D Vision hardware I can't say. If it doesn't work then there's not much I can do about it. I'm probably not going to add extra code for a proprietary NVidia solution. You could ask NVidia about it. I think they should be able to make it work for NVidia 3D Vision hardware. After all I'm using the official OS 3D APIs.

This is a known bug in the NVidia drivers. You can work around it by either using D3D9 presentation, or by telling madVR to enable/disable OS 3D support, depending on whether you play 2D or 3D content. Unfortunately NVidia 3D drivers are pretty shaky atm, especially in Windows 10.

Thanks! was just testing performance on my pc before transitioning over to the HTPC which uses standard 3D to the TV.

The D3D9 switch fixed the fullscreen crashing, thanks. The 3D switching was causing issues with the title bar of certain programs vanishing. Another Nvidia problem I would think.

aufkrawall

29th February 2016, 02:04

I have heard that before, so I did a quick test with some 1080i content that I typically watch. With a custom application setting of ""Prefer maximum performance", my rendering time is ~6 ms. When I change the setting to "Adaptive", the rending time goes up to 13 ms. Something is impacting the rendering times. My typical settings include ordered diffusion, Direct3D 11, Exclusive mode, and display at 1080i (video card is doing the scaling).

My display is natively 2560x1440p60. I played around with letting madVR do the scaling. I did the same test and obtained ~13 ms rendering times with either"Prefer maximum performance" or "Adaptive" power mode settings.

:confused:
The render times simply get lower when the clock is higher for the same task.
Sometimes the driver seems to decide differently about choosing what clock state for the same task, but this shouldn't really be "bugged behavior" as long as there are no frame drops.
However, I can't explain the weird behavior you described in the end. Maybe it's again unexpectable behavior, caused e.g. by DXVA deinterlacing?

ryrynz

29th February 2016, 02:22

Since v0.90.11 playback results in a black screen for me.

Looks to be the exact same issue I reported. He'll have it fixed in the next build.

jmelan

29th February 2016, 03:29

Well I sort of figured out what the problem with Nvidia 3d vision was:

I normally keep the monitor (asus pg278q) at 120hz

At 120Hz, 3d playback is enabled but I still see double with the glasses
At 60Hz, playback works. Is there a way to automatically switch to 60Hz when a 3D file is played?

...nevermind, the above only works occassionally

Switching to any resolution with the windows display properties option while the video is playing makes 3d work in windowed mode but causes intermittent crashing of MPC-HC when changing to fullscreen. Using the nvidia control panel to change anything seems less reliable. Seems more Nvidia bugs.

After some more testing, this only worked for few tries then stopped as well. After turning off stereoscopic 3d in the windows display properties and back on it started working again. no idea why the behavior is so inconsistent.

But the key to get it working was changing to a different 3D resolution or frequency while the video was already playing. After stopping the video and closing MPC-HC, and reopening the video, 3D again does not work until changing the resolution or frequency. Changing the resolution with the windows 10 display properties seems like the most reliable method.

Neeto

29th February 2016, 03:31

Tested. Don't see any difference at all. Present Queue drops the same with or without this option enabled when the seek bar is on screen in FSE mode. Also same stutter as before when seek bar is visible.

Didn't notice any sync problems.

Had a similar issue with queue draining very slowly and then not filling.
Had to clear both reclock database & change the vsync detection to the opposite I had set to get it stop it from happening.
Not sure if it's relevant - but just in case.

har3inger

29th February 2016, 07:35

I tried out the alternative glitch option on d3d11 playback (8 bit). My queues are now able to go above 2-3/3. (However, setting them to 16 will max them out at 14-15/15 for some reason).

I get presentation glitches "more", but only in situations that I'm supposed to have them, like when I empty the present queue on right clicking out of FSE, toggling full screen and FSE, seeking, etc. In normal playback, I don't get any extra presentation glitches.

I let a movie run for 5 minutes without pausing. No A/V desync that I could notice.

Thanks for the fix!

Werewolfy

29th February 2016, 10:56

When releasing v0.90.11 about a week ago I asked you guys for feedback about the following things:

1) Agressive downscaling anti-ringing filter: Is it working well for you, or do you see any problems?
2) New sharpening anti-ringing filter: Is it working well for you, or do you see any problems?
3) crispen edges: Do you prefer the linear light option to be enabled or disabled?

2) I really like it! I liked "sharpen edges" but there was a little bit of ringing sometimes that stopped me to always use it. Now it's very clean, I haven't seen any problems with it and I'm considering to use it with low values (and it's a big step for me because like I've said, I don't like sharpening filters).
3) Hard to tell... LL on is closer to the source (less bright with high sharpening values) but it generates artefacts around letters, I don't think it's ringing.
I will post screenshots ASAP but I still prefer LL off for now.

I don't like this look anymore either, but have you tried SuperRes against it? I think it works wonders, and unlike with super-xbr, haloing, ringing, aliasing and dark line boosting doesn't become an issue.
I tried various photographs and NNEDI3 + SuperRes always came closest to ground truth for me. And btw. even with a SR strength of 4, result was still much softer than ground truth.

So SuperRes can lessen the mosaic look that NNEDI3 produces on grass for example?
But SuperRes boosts the aliasing and ringing that NNEDI3 has hidden :(

FDisk80

29th February 2016, 11:52

Had a similar issue with queue draining very slowly and then not filling.
Had to clear both reclock database & change the vsync detection to the opposite I had set to get it stop it from happening.
Not sure if it's relevant - but just in case.

It's not really a problem unless you like watching movies with the seek bar on the bottom for some reason. :)

When it's hidden it's all fine.

aufkrawall

29th February 2016, 14:03

So SuperRes can lessen the mosaic look that NNEDI3 produces on grass for example?

Yep, SR makes it look more naturalistic.

But SuperRes boosts the aliasing and ringing that NNEDI3 has hidden :(
Seems we have to live with that. :(

leeperry

29th February 2016, 15:16

^ Yeah, really not convinced by the SR+NNEDI3 combo except on very low res.

aufkrawall

29th February 2016, 16:05

I'd be really interested to see an example where it actually fails in comparison to other settings in your opinion.

XMonarchY

29th February 2016, 17:21

Just a personal embarrassment story about my bug solutions with madVR.

My previous clean Windows 10 Pro x64 10240 RTM install on freshly formatted SSD resulted in madVR-rendered videos acting weird with top part of the screen being out-of-sync with the rest of the screen and insane number of drops + presentation glitches when videos were paused and then resumed. That happened with 4-6 builds of madVR and nobody else was experiencing this issue. No madVR settings adjustment helped either.

However, in every other way the system was rock-stable. No spyware, only 3 tiny background programs, no issues in games, no unnecessary services, a lot of crap disabled, all the latest drivers (DDU used before each install) and firmware, all the latest LAV filters, MPC-HC, all checked & re-checked, fully updated, used DISM RestoreHealth and SFC Scannow. It was flawless and yet madVR was messing up... Then I upgraded to Windows 10 x64 Pro 1511-10586 build on top of the original 10240 RTM build. Nothing changed.

I assumed it was madVR, but it wasn't because a few months later I clean-installed Windows 10 x64 Pro onto freshly formatted SSD simply because I wanted a clean start, not because there was anything wrong with my previous install. To my surprise, after 1511-10586 install, all madVR issues were gone. To this day, I get no madVR issues. I don't know what the problem was before, but it is gone now.

I am not saying you need to format and re-install Windows, but if you get an issue nobody else gets or one that is very uncommon, it might help to fresh-format and re-install OS, but I can't promise it will help.

Georgel

29th February 2016, 22:32

Anima123

1st March 2016, 01:20

Hello!

For me, with Nvidia default drivers, GTX860M laptop, I cannot use Direct3D11 without getting a mpc-hc crash. Everything works fine if I do not try to use Direct3D11

I will post crash report.

15301

It might be related to the crash reported I've submitted earlier, which madshi said it probably due to a bug in the driver.

ryrynz

1st March 2016, 02:05

It might be related to the crash reported I've submitted earlier, which madshi said it probably due to a bug in the driver.

Yeah the 860s haven't has a good run recently with Nvidia it seems. Try some earlier drivers.

x7007

1st March 2016, 05:08

huhn

1st March 2016, 06:17

If I have TV Philips 7007 LED/LCD Display IPS Panel , Which nit setting should I choose ? How can I know or test or what should I look for for correct ? Brightness : 450 cd/m2

What about my laptop display LG Display LP173WF4-SPD1 which is AHIPS I think , it says Brightness 300 , should that mean I need to choose 300 nit ? http://www.panelook.com/LP173WF4-SPD1_LG%20Display_17.3_LCM_overview_24463.html

you need a colorimeter to get such an info.

this info just means it could reach 450/300 cm² with maxed settings.

and this info is totally unreliable. over time displays will lose max brightness too.

not sure if i would even try to increase brightness on an IPS screen.

edit: set the HDR nits to that that looks best to you.

har3inger

1st March 2016, 06:41

Georgel

1st March 2016, 11:33

Yeah the 860s haven't has a good run recently with Nvidia it seems. Try some earlier drivers.

I actually tried all drivers versions that would work with Windows 10. It seems that nothing so far would repair the situation, and I did send more than a few reports to nvidia about driver problems.

Georgel

1st March 2016, 11:35

If I have TV Philips 7007 LED/LCD Display IPS Panel , Which nit setting should I choose ? How can I know or test or what should I look for for correct ? Brightness : 450 cd/m2

What about my laptop display LG Display LP173WF4-SPD1 which is AHIPS I think , it says Brightness 300 , should that mean I need to choose 300 nit ? http://www.panelook.com/LP173WF4-SPD1_LG%20Display_17.3_LCM_overview_24463.html

The laptop display has been confirmed to be around 300 nits by notebookcheck and laptopmedia.

I still do not understand exactly what hdr does, but as soon as I get my hands on some HDR sample videos, I will test it out.