madVR with LG OLED C8

[Asmodian]

There really is no conclusive test for 10-bit. I came to that realization years ago, when we first started that thread here on doom9. Dithering will always throw things off. People won't stop trying, though.

Yes, we can only test entire playback chains.

Discovering (and accepting) this truth is what broke my quest for a real 10 bit display chain too. If I cannot reliably figure out if the playback chain is dithering why is madVR's tunable high quality dithering a bad thing?

For the C8 (and C7 and C9) madVR even does a much better job than the display, so the desire for a 10 bit playback chain ends up hurting quality instead of helping. :o

[huhn]

I'd certainly call RTINGS trustworthy, and they do some of the most detailed testing, but still they don't scrutinize enough. Not the way that some of us here would do.

they are the best source we have access to but that doesn't change that i think there banding test is flawed.

@ThunderFi
you have to disable d3d11 and FSE or overlay will be disabled when you enter fullscreen.

[VBB]

Thunder, if it only stutters in full-screen, do you mean full-screen exclusive or even windowed full-screen? FSE basically doesn't exist in the newer versions of Windows 10, so setting madVR to windowed FS is probably the best option. Don't know if that will help your situation, though.

[j82k]

Rtings gradient score seems to be completely useless. My previous TV was the LG C6 and my current one is the C8. The most noticeable improvement for me was the smoother gradient handling of the C8, especially with darker colors.

Now looking at the rtings score:
LG C6: Gradient 9.5 :eek:
LG C8: Gradient 9.1 :confused:

edit: it's even visible on their photo, the C6 shows some really nasty bands in the darker colors, how the hell did they give it a 9.5??

[VBB]

Granted, though they did have a different scoring system when the C6 came out. Can't compare them directly.

[j82k]

Ok didn't really pay attention to that... but still the C6 was terrible at gradients...

[ThunderFi]

Thunder, if it only stutters in full-screen, do you mean full-screen exclusive or even windowed full-screen? FSE basically doesn't exist in the newer versions of Windows 10, so setting madVR to windowed FS is probably the best option. Don't know if that will help your situation, though.

Sorry for the confusion. I just meant if the program is windowed vs fullscreen. I wasn't talking about the madVR options. By windowed FS do you mean windowed overlay? That's the only way to have MPC-HC in fullscreen without stuttering, of course I still need to disable D3D11. But I thought that could be a lead to the issue. If the video works without stuttering when MPC-HC is windowed but the problem occurs once it's fullscreen. Any ideas?

EDIT: And when I say fullscreen I'm not talking about FSE.

[huhn]

another reasons could be that your settings are to doable with your GPU in real time.

[ThunderFi]

another reasons could be that your settings are to doable with your GPU in real time.

Wait a minute. Do you mean my settings are too heavy for my GPU or the other way around?

My frame times on average are about 27 ms and the 5s maximum rarely goes to even 40 ms. I have no frame drops or repeats when the stuttering happens.

[VBB]

huhn, he has a 2080, so that's unlikely to be the issue.

[VBB]

Thunder, you tried switching between FS exclusive and FS windowed by unchecking/checking the box in madVR? Windowed overlay is an entirely different option.

[ThunderFi]

Thunder, you tried switching between FS exclusive and FS windowed by unchecking/checking the box in madVR? Windowed overlay is an entirely different option.

Yes, I have also tried this.

Right now I'm experiencing a weird bug (maybe). When I drag and drop MPC-HC into my TV with an HDR movie the TV recognizes the HDR signal. When I put the player into full screen it still says HDR but it reverts to SDR and the picture just looks washed out. The only way this doesn't happen with D3D9 is if I use windowed overlay. With D3D11 it doesn't matter if I have FS exclusive or FS windowed or windowed overlay, the HDR works as it's supposed to. Is the problem caused because I still have D3D11 selected in MPH-HC?

Anyway, I noticed that when I play a movie with VLC the stuttering is a bit different than with MPC-HC + madVR but it's still there. It seems to be more subtle with VLC but still noticeable.

EDIT: Changing D3D11 in MPC-HC to DXVA2 (copy-back) didn't help with the bug or whatever I'm experiencing.

[VBB]

Just to be sure, you are using the latest available MPC-HC/LAV, madVR, and Nvidia drivers? There were some older drivers where it wouldn't trigger HDR unless you selected 10-bit output in madVR, but that has been fixed since. I wish I had a RTX myself, so I could help you better, but I'm waiting for HDMI 2.1 support before I buy another video card.

[QBhd]

Is there an option in your player to open the video on the TV even though the player is on the monitor? Moving video windows around is always problematic. I have PotPlayer on my monitor, but soon as I hit play, it opens the video full screen on my C8 OLED.

QB

[ThunderFi]

Just to be sure, you are using the latest available MPC-HC/LAV, madVR, and Nvidia drivers? There were some older drivers where it wouldn't trigger HDR unless you selected 10-bit output in madVR, but that has been fixed since. I wish I had a RTX myself, so I could help you better, but I'm waiting for HDMI 2.1 support before I buy another video card.

Yes. My MPC-HC is 1.8.7. My madVR is 0.92.17. My NVIDIA driver is 436.02. Actually a new NVIDIA driver was released today, I might try that tomorrow since it's getting late here and I'm soon going to sleep.

The HDR "bug" is only with D3D9, with D3D11 the HDR works fine with every setting but the video stutters.

Is there an option in your player to open the video on the TV even though the player is on the monitor? Moving video windows around is always problematic. I have PotPlayer on my monitor, but soon as I hit play, it opens the video full screen on my C8 OLED.
QB

Yes there is. I just tried it now. It didn't help, the same problem persists.

Anyone, this is a problem that I feel like is futile to solve. This problem gets fixed when I use D3D11 but I can't use it because of the stuttering. It should work, I just need to find a way to fix it. Even if I can't fix it I can still use D3D9 with windowed overlay and everything is fine.

EDIT: I haven't updated any LAV's myself. I thought the latest MPC-HC would have the newest ones or at least ones that are new enough. If I had outdated LAV's would that affect VLC as well then?

[VBB]

With D3D9, you want to use overlay anyway, so you should be good for now.

[ThunderFi]

With D3D9, you want to use overlay anyway, so you should be good for now.

Okay that's good to know. Yeah for sure. It works just as before but I think the HDR "bug" is a new thing, at least I hadn't noticed it before.

Right now this is a waiting and googling game for me. I feel like I've googled "everything" even though that's just impossible. I have tried everything that comes to mind. The only new thing left to do right now is to try the newest drivers that were released today. If that doesn't help then I don't know what to do. I have tried disabling GPU overclocks. Maybe I'll try disabling CPU overclocks as well.

EDIT: The thing that is most weird to me is the fact that there is no problems with my other two monitors, only with the LG C8.

[ThunderFi]

I want to say that I appreciate everyone very much who's come here and tried to help me. It really means a lot to me. Seeing people who know so much treat me, a guy who knows very little about these things, in such a warm and kind way. Thank you very much everyone.

I hope we can get this fixed eventually.

Also, just wanted to say that I'm sorry if I seem cold or maybe even mean (hopefully not). I'm dealing with some difficult stuff IRL and that affects me negatively. Trying to stay positive to tackle this problem though!

[VBB]

I was just reading the update notes for the new Nvidia driver when I came across this:

"Windowed NVIDIA G-SYNC is Disabled During Video Playback
on a Multi-monitor System

Issue

Beginning with Windows 10 April 2018 Update (RS4), NVIDIA G-SYNC is disabled
during active video playback on multi-monitor systems.

Explanation

If hardware accelerated video is playing on one display and an application is launched on
another display in windowed mode with NVIDIA G-SYNC or adaptive-sync enabled, the
video stutters. This is the result of desktop compositing on displays with differing refresh
rates.
To prevent the video stutter in this scenario, the driver disables NVIDIA G-SYNC during
active video playback, but re-enables G-SYNC if the video is paused or closed.

Note: The issue can also occur when displays are set to different
refresh rates, even when G-SYNC is not used. To resolve in this case,
make sure all displays are set to the same refresh rate."

Could this be the problem? Have you tried setting all your monitors to the same refresh rate as your TV? I forgot if you mentioned this before, but does the stuttering occur when you have only the TV connected?

[huhn]

the desktop composition is pretty broken on windows 10 but overlay should workaround that.

the biggest problem currently is with RTX cards and some user can't get them to work properly even with overlay and some other user have no issues at all.

[VBB]

Well, that's why I'm thinking this might be his exact issue, because he does use two other monitors with different refresh rates along with the C8, if I remember correctly. Would explain why overlay works for him, but anything else doesn't.

[ThunderFi]

I was just reading the update notes for the new Nvidia driver when I came across this:

"Windowed NVIDIA G-SYNC is Disabled During Video Playback
on a Multi-monitor System

Issue

Beginning with Windows 10 April 2018 Update (RS4), NVIDIA G-SYNC is disabled
during active video playback on multi-monitor systems.

Explanation

If hardware accelerated video is playing on one display and an application is launched on
another display in windowed mode with NVIDIA G-SYNC or adaptive-sync enabled, the
video stutters. This is the result of desktop compositing on displays with differing refresh
rates.
To prevent the video stutter in this scenario, the driver disables NVIDIA G-SYNC during
active video playback, but re-enables G-SYNC if the video is paused or closed.

Note: The issue can also occur when displays are set to different
refresh rates, even when G-SYNC is not used. To resolve in this case,
make sure all displays are set to the same refresh rate."

Could this be the problem? Have you tried setting all your monitors to the same refresh rate as your TV? I forgot if you mentioned this before, but does the stuttering occur when you have only the TV connected?

I had not tried to set all of my displays to the same refresh rate. I just tried it, I set all of them to 60 Hz. Didn't set them to 23 Hz because my other two displays don't have 23 Hz by default. Like I mentioned previously at some point, the higher my TV's refresh rate is the less noticeable the stuttering is. So at 60 Hz the stuttering turns into like a micro stutter that makes the playback seem "smooth" but it really isn't, if that makes sense. If I put my TV back into 23 Hz the stuttering is much more noticeable.

I had also tried to only have my TV connected but that didn't help. I tried it again just to be sure. I disabled my other two monitors and restarted my PC. Having only my TV active didn't help unfortunately. I tried it with 23 Hz and with 60 Hz and what I said above was the same when only the TV was connected.

Well, that's why I'm thinking this might be his exact issue, because he does use two other monitors with different refresh rates along with the C8, if I remember correctly. Would explain why overlay works for him, but anything else doesn't.

I wish it was this because then we would know what's causing the issue, but it would seem that this is not the case :(

EDIT 1: Someone posted their madVR GPU flushing settings on reddit to me and those seem to work even with D3D11. If I change my windowed and exclusive mode settings to "don't flush" and only 1 frame presented in advance there's no stutter. I did this to both windowed and exclusive mode. With this I can use windowed FS without problems, also FSE works as well, haven't tried windowed overlay yet. Though this is still a fix just for using MPC-HC + madVR. VLC for example still stutters. Can we draw some conclusions from this? Do other media players flush the GPU in a way that causes it to stutter? How can the default settings be wrong for such an "advanced" GPU?

EDIT 2: Those settings cause a lot of presentation glitches on my main monitor. I would love to get a working video playback on all of my screens so even this is a temporary fix. I think using D3D9 with windowed overlay is a better temporary fix.

[ThunderFi]

One more thing I've noticed:

SDR movies:
- VLC = no stutter
- MPC-HC + madVR = stutter

HDR movies:
- VLC = stutter (HDR doesn't even work with my VLC apparently)
- MPC-HC + madVR = stutter

EDIT: And these apply when using my TV.

I don't know if this helps but I thought to post some extra info if someone might think of something. These results apply with D3D11. Again, D3D9 with windowed overlay work well in MPC-HC + madVR, doesn't matter if the movie is SDR or HDR.

I also tried resetting MPC-HC to default settings and also installed madVR again, neither of these helped.

[chros]

Have you seen this post (https://www.avsforum.com/forum/139-display-calibration/3088376-lg-oled-2019-pq-curve-upload-free-template-devicecontrol-interface.html#post58608526) by Ted? It's about tonemapping on C9, but it's supposed to be the same on 8 series as well.

The interesting parts are:
- deafult peak nits is: 700
- the rolloff value of the lowest curve is 70%
-- that means 490 nits! where the rolloff starts (probably that's why we saw clipping previously)

Now, the question is why madvr needs to be set 760 nits to trigger the lowest curve and not 1000 or 990?

@quietvoid, can you upload the sm test (https://www.avsforum.com/forum/37-video-processors/3053602-madvr-envy-anticipation-thread-48.html#post58598174) file somewhere? Thanks

[quietvoid]

madVR test build v91 seems to fix the dynamic tonemapping for my LG C8.
Here's a comparison between v86 static 760 nits vs. v91 with dynamic tonemapping: https://slowpics.org/comparison/e38b56e5-7212-464c-8e45-27dd8f98d701

[Warner306]

That is a dramatic difference. Are you referring to madVR's dynamic tone mapping in output video in HDR format with the display set to do static tone mapping?

[iSeries]

madVR test build v91 seems to fix the dynamic tonemapping for my LG C8.
Here's a comparison between v86 static 760 nits vs. v91 with dynamic tonemapping: https://slowpics.org/comparison/e38b56e5-7212-464c-8e45-27dd8f98d701

Yes difference is huge there. Could you link to a screenshot of your madVR tonemapping options screen with the settings used to achieve this please?

[quietvoid]

That is a dramatic difference. Are you referring to madVR's dynamic tone mapping in output video in HDR format with the display set to do static tone mapping?
Yes this is with output as HDR, the TV has LG's DTM enabled.
There's little difference for highlights with LG DTM ON, however low nits scenes are brighter than they should be.

Yes difference is huge there. Could you link to a screenshot of your madVR tonemapping options screen with the settings used to achieve this please?
These are my settings
https://i.ibb.co/5Tf69yc/madvr.png (https://ibb.co/5Tf69yc)

[Warner306]

Does it work as well with LG's dynamic tone mapping disabled?

If it does work well with static tone mapping, would it be possible for you to make 2-3 screenshot comparisons of madVR's dynamic tone mapping vs. LG's static tone mapping and post a link here? The horses in the snow is already one excellent comparison.

I'd like to add those photos to my madVR guide as I haven't collected any screenshots for HDR passthrough, only HDR to SDR.

[chros]

If it does work well with static tone mapping
I think I found another bug with passthrough (https://www.avsforum.com/forum/24-digital-hi-end-projectors-3-000-usd-msrp/2954506-improving-madvr-hdr-sdr-mapping-projector-215.html#post58647476), can someone confirm this if you have a C8/E8 or B9/C9/E9 by looking at the 2nd last point of 20 point WB value in LG menu?

[quietvoid]

Does it work as well with LG's dynamic tone mapping disabled?

If it does work well with static tone mapping, would it be possible for you to make 2-3 screenshot comparisons of madVR's dynamic tone mapping vs. LG's static tone mapping and post a link here? The horses in the snow is already one excellent comparison.

I'd like to add those photos to my madVR guide as I haven't collected any screenshots for HDR passthrough, only HDR to SDR.

I've made this comparison for you, hopefully it is useful.
https://slowpics.org/comparison/1b0b83e2-8a5e-43cf-8de8-4c6ab4840c98

To explain what I see:
madVR DTM without LG DTM looks better than passthrough without LG DTM.

When it comes to enabling LG DTM, passthrough is going to look similar compared to madVR + LG DTM OFF.

And madVR + LG DTM ON looks slightly more detailed, but then you get inconsistent brightness changes for everything.

So if you don't mind the brightness fuckery LG does with the DTM, you should be fine with passthrough.
Otherwise, madVR without LG DTM is better.

For myself I will be changing to madVR dynamic tonemapping alone, disabling LG's DTM. :)

I think I found another bug with passthrough (https://www.avsforum.com/forum/24-digital-hi-end-projectors-3-000-usd-msrp/2954506-improving-madvr-hdr-sdr-mapping-projector-215.html#post58647476), can someone confirm this if you have a C8/E8 or B9/C9/E9 by looking at the 2nd last point of 20 point WB value in LG menu? With the Spears & Munsil disc (10,000 nits), passthrough ends up setting the WB point to 713. I'll need to try other titles to see.

[chros]

With the Spears & Munsil disc (10,000 nits), passthrough ends up setting the WB point to 713. I'll need to try other titles to see.
That sm clip is not a good test for that because both master luminance and maxcll are 10000.
The mockingjay clip is a good one, maxcll is around 680 but master luminance is 1000, so the middle tone curve "should be" used for now.

And I switched to madvr dtm + LG dtm :) (I set up the Cinema Home preset to be the same as Cinema apart from LG DTM, so you can quickly switch between them. I think Color temperature should be set to W50 in Cinema Home to get Warm2 in Cinema).

[Warner306]

I've made this comparison for you, hopefully it is useful.
https://slowpics.org/comparison/1b0b83e2-8a5e-43cf-8de8-4c6ab4840c98


Thanks for the images. Appreciate it.

Can you also take two more screenshots: One with madVR's DTM enabled and one with LG's static tone mapping alone with the horses in the snow scene? I really like that example.

[quietvoid]

Can you also take two more screenshots: One with madVR's DTM enabled and one with LG's static tone mapping alone with the horses in the snow scene? I really like that example.
Here you go:
Passthrough vs. madVR.
https://i.ibb.co/bz1v7CW/passthrough.png (https://ibb.co/bz1v7CW) https://i.ibb.co/dcV25hJ/madvr.png (https://ibb.co/dcV25hJ)

Both have LG's dynamic tonemapping disabled.
Keep in mind that sample is very bright, mastered at 10,000 nits.
And the display is probably not outputting over 300 nits because it is almost full field.
It does not represent most of the available content out there, for which there might be much less difference.

And these are the settings, for test build v91:
https://i.ibb.co/ssYhsmy/settings.png (https://ibb.co/ssYhsmy)

[Warner306]

Thanks.

[shae]

The previous thread is not enough? :)

https://forum.doom9.org/showthread.php?t=176215

[ThunderFi]

Hey! I'm the OP of this thread.

I had given up on getting my setup working perfectly after a lot of failures upon failures, that's why I haven't been here in a while. I haven't yet tried reinstalling Windows (which is the only thing I have left to do). Right now I'm using NVIDIA CP default 23 Hz mode when watching movies + 8 bit RGB full. It's far from perfect but it's okay.

I'm kind of worried of trying to create custom resolutions with CRU again. If I do something wrong can it lead to having to reinstall Windows again?

[oldpainlesskodi]

Hey, flip the to red side, cus they are so much better....not. I think it's a trade off at this stage. AMD (pretty much perfect 23p, and you can control the colour temp (64k white point), Nivida = more stable with HDR (no control over white point cus there is no edid override.etc).

Everyone gets choose - green team, or red team....facepalm moment right there.

[huhn]

what has this to do with the white point the last thing you want from the GPU driver is changing the white point of your processed and dithered image.

[oldpainlesskodi]

Wow, I expected you to get it. With NV i have no control, Amd I do ( I can override the edid...all film content is 64k...unless I'm about to get hit with a frying pan upside the head (oldpainess ducks just in case) :-)

Just for fun....try setting AMD Temperature control Temperature control from Auto to Manual.

[huhn]

the spec for bt709 has a white point of D65 that's the white point you get if you don't touch the white point.

and that's again irrelevant if your screen is not calibrated because it will have a white point of "what ever" usually in the d70-d80 range.

you can do the same bad adjustment in nvidia but not like amd with only one option you have all 3 RGB channels to adjust the white point and it should be avoided for the same reason.

if your screen is calibrated to D65 this option may do something but why you would do it that way is beyond me it will lower your
CR.

[nevcairiel]

As a general rule, stay away from any color adjustment in the graphics cards control panel. Its generally a low quality algorithm. If you really need to change something, do it with madVR.
Not sure what "EDID Override" has to do with that anyway. And with tools like CRU you can make any EDID for any graphics card.

[oldpainlesskodi]

Yeah, I think you are right - been digging deeper.

[chros]

Nivida ... there is no edid override.etc).
There is (https://forum.doom9.org/showthread.php?p=1902402#post1902402) ...

[oldpainlesskodi]

Sorry, I meant in terms of temperature, well, not that I can remember...