Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Hardware & Software > Software players

Reply
 
Thread Tools Search this Thread Display Modes
Old 27th September 2013, 22:20   #20141  |  Link
Plutotype
Registered User
 
Join Date: Apr 2010
Posts: 234
my friend copied these setting onto his setup:
http://www.mediafire.com/?y8x1s1u8et6s92w
__________________
__________________
System: Intel Core i5-6500, 8GB RAM, GTX960/GTX1060, 55" Sony Bravia KDL-55HX850 @ 1920x1080 24p, Yamaha YSR-1100 surround bar
Setup: Win7/Win10 Pro, madvr/JRiver 24
Plutotype is offline   Reply With Quote
Old 27th September 2013, 22:27   #20142  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,249
Try disabling 'use separate device' also try disabling 'switch to matching display mode'
Also, unrelated to this issue, untick 'scale in linear light' on upscaling.
ryrynz is online now   Reply With Quote
Old 27th September 2013, 22:34   #20143  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 9,860
Are you sure ReClock is actually doing its job properly?
Because it still says that it'll drop a frame every 30 seconds, if ReClock is functioning perfectly, that number should go way down to hours if not days.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 27th September 2013, 23:18   #20144  |  Link
Qaq
AV heretic
 
Join Date: Nov 2009
Posts: 422
Check bitstreaming in Lav audio.
Qaq is offline   Reply With Quote
Old 28th September 2013, 07:43   #20145  |  Link
dansrfe
Registered User
 
Join Date: Jan 2009
Posts: 1,212
Is there a reason why the "1 frame repeat/drop ever X units" may be hours at one point and days at another? What actually determines "X units"? How does ReClock affect this value, if the refresh rate is a multiple of the frame rate and the taskbar icon is green?
dansrfe is offline   Reply With Quote
Old 28th September 2013, 09:44   #20146  |  Link
Plutotype
Registered User
 
Join Date: Apr 2010
Posts: 234
Thanks guys, the Reclock is set correctly ( lightning green ) according to this http://img33.imageshack.us/img33/1411/ybhj.png
He says, he is loosing around 25 frames in 3 minutes of a 1080p video and the drops are apparent in fast moving scenes. It doesnt look like a madVR issue to me, so I wont spam this thread and try to fix it by checking his PC state and other components during playback.
__________________
__________________
System: Intel Core i5-6500, 8GB RAM, GTX960/GTX1060, 55" Sony Bravia KDL-55HX850 @ 1920x1080 24p, Yamaha YSR-1100 surround bar
Setup: Win7/Win10 Pro, madvr/JRiver 24

Last edited by Plutotype; 28th September 2013 at 10:07.
Plutotype is offline   Reply With Quote
Old 28th September 2013, 12:40   #20147  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,137
Quote:
Originally Posted by 6233638 View Post
Something I have noticed for a while, but never had the time to chase down the cause of, is that black level is always slightly raised when using madVR - it would never fade to black, and I had to reduce brightness by one or two steps further than should be necessary, sacrificing some shadow detail.

It turns out that the cause is madVR's dithering.
Would it be possible to disable dithering for black, or maybe even the lowest 1% of the signal, rather than having to disable dithering entirely?
madVR processes in high bitdepth. At the end of the processing chain, some pixels might have 8bit values of 0.1/0.1/0.1, while others have 0.0/0.0/0.0. If I don't apply dither to the 0.0/0.0/0.0 pixels, quantization noise might creep back in and damage image quality in dark scenes.

It is true that dithering hurts black levels (and white level, for that matter) a tiny bit, but at the moment there's nothing I can do about that. Oh well, I could detect a full black/white frame and behave differently, but that would cost quite some additional performance. In the long run I hope to be able to replace dithering with error diffusion. That should take care of the problem. However, this will only be possible (if at all) by using OpenCL or CUDA, and that is still some time away...

Quote:
Originally Posted by Plutotype View Post
Do I need to restart the PC/notebook when changing levels using madNvLevelsTweaker? Has Nvidia addressed the RGB Full/Limited levels issue ( missing full-limited instant selection )? I dont have a Nvidia GPU to test it, sorry.
Yes, a reboot is necessary, and no NVidia hasn't changed anything, AFAIK.

Quote:
Originally Posted by Plutotype View Post
Other question I need to get answer is - how much GPU intensive is 3D LUT processing on your GPU using madVR? Will this be a non issue on a Intel HD4400 and a i7 Haswell CPU?
It costs performance, but it works well enough on my HD4000 with 1080p24 movies. Haven't tested 1080p60 playback with 3dlut.

Quote:
Originally Posted by cvrkuth View Post
I update my OS (Windows 8.1 Pro), I have same problem again.
It works for me. Which GPU do you have? Are you using the latest madVR build? If so, create a debug log, zip it and upload it somewhere and I'll have a look.

Quote:
Originally Posted by StephaneM View Post
I already placed the drain on the video window host (with or without the drain, there is no changes).
You need to place the drain on madVR. And the thread which owns the window which you've used for the drain must be quick to react to changes, if you don't want delays in mouse/keyboard handling.

Quote:
Originally Posted by Gagorian View Post
It seems the same issue (the lines/blurring close to the picture frame border) is also visible when using EVR. It's present on both my computer (AMD HD 7950) and my projector (HTPC with AMD A10-5700 iGPU).

It seems that it isn't a madVR specific problem, but do you have any ideas what could be causing it?
Quote:
Originally Posted by 6233638 View Post
It will be the source material. A lot of SD content has this problem, and even some HD content.
I agree with 6233638, I think these ringing artifacts are most likely already in the source. The only thing you could do is use avisynth processing to try to remove the artifacts. avisynth is not too easy to setup for an avisynth unexperienced user, unfortunately.

Quote:
Originally Posted by Plutotype View Post
Hi guys,
a friend of mine cant get his madVR/MPC-HC/LAV/reclock setup right. He has an Core i7 desktop CPU, GTX780 GPU and gets this:
http://img585.imageshack.us/img585/2907/0uxn.jpg

We cant track down the reasons for the dropped frames. What else would you check?
It's weird that you have both dropped and repeated frames, and so many of them. I'd suggest to try resetting Reclock's timing database. If that doesn't help, try without Reclock. If that still doesn't help, you could upload a debug log for me to look at. You're using LAV for splitting & decoding, I suppose?

Also make sure you have no other software running which might use the GPU. Might even be the browser.

-------

Next madVR build is going to have some goodies for you guys. Will take a while until it will be released, though. Here's one new feature:

http://forum.doom9.org/showthread.php?p=1645630#post1645630
madshi is offline   Reply With Quote
Old 28th September 2013, 12:48   #20148  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Is there any way to use ICC Color Profiles (like MPC-HC does natively) without converting to 3D LUT?

I tried to convert my profile using Argyll+DispcalGUI, and with TI3 Parser but with no success,
both of them show some crazy windows errors with red X in them.


I calibrate my monitor (Dell U2410 in sRGB mode) with i1 Display Pro to create an ICC profile.

In madVR in the Calibration settings:
I see NO difference between "Disable calibration controls for this display"
and "This display is already calibrated" with BT.709, pure power curve 2.20.
Gamma Processing works just the same.


The problem is when I switch to a wider gamut option in my display I have no way to use my calibration profile.
Again, Is there any way to use ICC Color Profiles (like MPC-HC does natively) without converting to 3D LUT?
James Freeman is offline   Reply With Quote
Old 28th September 2013, 13:06   #20149  |  Link
StephaneM
Registered User
 
Join Date: Jun 2006
Posts: 15
Quote:
Originally Posted by madshi View Post
You need to place the drain on madVR. And the thread which owns the window which you've used for the drain must be quick to react to changes, if you don't want delays in mouse/keyboard handling.
I'm not sure I know how to put the drain on madVR ?

This is what I'm doing for now :

Code:
playbackGraph.AddFilter(CommonFilter.MadshiVideoRenderer, "evr");
.../... (Building Graph : adding filters connecting pins)
IVideoWindow videoWindow = playbackGraph.GraphBuilder.GetInterface<IVideoWindow>();
result = videoWindow.put_Owner(videoFrame.VideoHwnd);
result = videoWindow.put_MessageDrain(videoFrame.VideoHwnd);
result = videoWindow.SetWindowPosition(0, 0, (int)videoFrame.WindowSize.Width, (int)videoFrame.WindowSize.Height);
result = videoWindow.put_WindowStyle(DSWindowStyle.Child | DSWindowStyle.Visible | DSWindowStyle.ClipSiblings);
The "videoFrame" is a WPF Control which hosts the WindowsFormsHost, this host does have a Window, this is this window handle that is exposed through videoFrame.VideoHwnd and forwarded to madVR (with put_Owner)

This window is created on the WPF UI thread (I have no other choices), while madVR is created on a dedicated thread (STA or MTA, with message pumping)

Quote:
Originally Posted by madshi View Post
Next madVR build is going to have some goodies for you guys
This debanding feature is very nice. Thanks.
StephaneM is offline   Reply With Quote
Old 28th September 2013, 13:17   #20150  |  Link
pie1394
Registered User
 
Join Date: May 2009
Posts: 212
Quote:
Originally Posted by 6233638 View Post
It's an HX900.

When dither is enabled, a black signal is displayed as a mixture of 1 and 0, which activates the backlight at a very low level. It's still very dark, and was difficult to photograph at all, which is why I have the camera at a sharp angle to the TV like that. Any photos I took looking straight on at the TV were just black.

If you're in a dark room and sit off to the side of the TV, you should be able to see it with the 920 as well, unless they have changed their local dimming algorithms, allowing it to clip the lowest signal levels. Or it may be that you have brightness set to a level which clips the lowest levels.
I guess your eyes are exteremely good.

Anyway I did a simple test and 3 images were taken for the comparison. The 1st one is regular lighting room condition. The 2nd one is the paused video at totally dark scene with mouse cursor moving around. The 3rd one is the same video w/o mouse cursor. The latter 2 photos were taken with ISO100 / 30sec / F1.4 setting. The camera is Panasonic GF2 + Leica DG 25mm.

The GPU is HD7970 with 1920x1080p24 10-bit color mode signal to 65HX920. Video source is BD, MadVR setting is with Jinc3 AR, dithering ON, FSE. But I don't know if the GPU driver's 8-bit to 10-bit LUT (or even dithering?) could lead the different result... ?

Yet it is confirmed LEDs will be shut-off entirely at the dark-zones of the 2nd one - just like the 3rd attached image.

The 2nd one does show some light bleeding (more obvious at bottom of the panel) while it is not seen on the 3rd one. It can also observed by my eyes observing from the same off-angle. But I think it is the panel's design characteristics -- might be diffuser film related limitation.

Anyway 65HX920's Innolux S-MVA panel, local dimming control, Corning Gorilla class seems to perform better than HX900's Sharp panel + glass (according to some reviews).
Attached Images
   
pie1394 is offline   Reply With Quote
Old 28th September 2013, 18:35   #20151  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by madshi View Post
madVR processes in high bitdepth. At the end of the processing chain, some pixels might have 8bit values of 0.1/0.1/0.1, while others have 0.0/0.0/0.0. If I don't apply dither to the 0.0/0.0/0.0 pixels, quantization noise might creep back in and damage image quality in dark scenes.

It is true that dithering hurts black levels (and white level, for that matter) a tiny bit, but at the moment there's nothing I can do about that. Oh well, I could detect a full black/white frame and behave differently, but that would cost quite some additional performance. In the long run I hope to be able to replace dithering with error diffusion. That should take care of the problem. However, this will only be possible (if at all) by using OpenCL or CUDA, and that is still some time away...
I suppose I'll just have to disable dither for now then. It may not matter on panels with with less than perfect black level, but on a local dimming LED set, my only option is to reduce brightness and clip the lowest levels, or disable dithering and retain that shadow detail. It's the difference between a zone switching off, and always staying on at a low level.

Quote:
Originally Posted by madshi View Post
Next madVR build is going to have some goodies for you guys. Will take a while until it will be released, though. Here's one new feature:

http://forum.doom9.org/showthread.php?p=1645630#post1645630
Looks promising! I assume there will be a new tag so it can be selectively enabled/disabled?
6233638 is offline   Reply With Quote
Old 28th September 2013, 19:50   #20152  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Anyone use the BT.709 Gamma profiles, and think they look like they got elevated blacks and too low gamma (1.95)?

That's because 709 gamma specs are for Encoding ONLY !!!
NOT for calibrating displays, period.

The film you are watching or a recording from a video camera is probably is already encoded with a standard 709 gamma.
If its a professional Hollywood film aka (Blu Ray), its 100% encoded with rec.709 gamma curve.
The Rec.709 gamma curve is encoded into the Camera while filming (or converted to in the studio).

Your Display calibration should be a Pure Power Curve of 2.2 to 2.4.

Again,

The Rec.709 Standard refers to Encoding Only.
The resulting video should be watched on a Pure Power Curve gamma display like a CRT or LCD.


Why you ask?
The 709 encoding gamma curve is designed to reduce the blacks below 1.8% of luminance to remove digital Camera noise.
When we use the Inverse curve for calibrating displays (which we absolutely should NOT), we get elevated blacks below 1.8%, a gamma of 1.96, and a lower Contrast Ratio.
When viewing a 709 encoded video on a properly calibrated monitor with a power curve of 2.2 to 2.4, these blacks (below 1.8%) are blacker than
an encoded video with a pure power curve gamma of 0.45 or 0.50 (like with older SD NTSC standard).
All this to eliminate the noise at the lower blacks 6233638 saw few posts ago, although not because of the same reason.

Everything falls into place for our enjoyment of video watching.


Quote:
While there may be several problems here, you're getting a few things confused. Rec.709's "overall gamma of 1.9" is in the encoding curve.
A problem with rec.709 has been that the spec only includes the encoding curve (what the camera uses to translate linear light into the spec's color space).
This curve is not relevant for monitors. You'll want to calibrate your monitor somewhere between Gamma 2.2 and 2.4 to display video correctly.
When you say "sRGB Gamma 2.2", I assume you're referring to sRGB primaries and a Gamma 2.2 curve, which is correct, since sRGB and rec.709 primaries are identical.
Quote:
Again, if your calibration software is offering "gamma 2.2" and "gamma 2.4" as an alternative to "rec.709", then they're referring to the gamma curve, and you should use 2.2 or 2.4.
Many calibration systems sadly don't understand that the rec.709 curve is not suitable for anything but encoding and should not be given even as an option for monitor calibration.

Question: Why madshi included BT.709 Display Gamma into MadVR?



Hope this helps to clarify my confusion.

Last edited by James Freeman; 28th September 2013 at 20:25.
James Freeman is offline   Reply With Quote
Old 28th September 2013, 20:16   #20153  |  Link
PetitDragon
Registered User
 
Join Date: Sep 2006
Posts: 80
Quote:
Originally Posted by James Freeman View Post
....If its a professional Hollywood film aka (DVD, Blu Ray), its 100% encoded with rec.709 gamma curve....
Are you sure? Genius!

PetitDragon is offline   Reply With Quote
Old 28th September 2013, 20:18   #20154  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Quote:
Originally Posted by PetitDragon View Post
Are you sure? Genius!

No, I'll fix it.

It may very well be Rec.601, with encoded power curve of 0.45 in the camera.
For an End-to-End gamma of 1.125 (0.45 *CRT Gamma 2.5).

Last edited by James Freeman; 28th September 2013 at 20:25.
James Freeman is offline   Reply With Quote
Old 28th September 2013, 21:13   #20155  |  Link
e-t172
Registered User
 
Join Date: Jan 2008
Posts: 568
Quote:
Originally Posted by James Freeman View Post
Is there any way to use ICC Color Profiles (like MPC-HC does natively) without converting to 3D LUT?

I tried to convert my profile using Argyll+DispcalGUI, and with TI3 Parser but with no success,
both of them show some crazy windows errors with red X in them.
No, but Argyll is now able to generate 3DLUTs directly with great flexibility. See the Argyll doc and this thread for instructions.

Quote:
Originally Posted by James Freeman View Post
Anyone use the BT.709 Gamma profiles, and think they look like they got elevated blacks and too low gamma (1.95)?

That's because 709 gamma specs are for Encoding ONLY !!!
NOT for calibrating displays, period.
That's correct. ITU-R BT.709 only defines an opto-electronic transfer function, as opposed to an electro-optical transfer function.

Quote:
Originally Posted by James Freeman View Post
Your Display calibration should be a Pure Power Curve of 2.2 to 2.4.
Not quite. Ideally it should follow the ITU-R BT.1886 specification, which is the new standard for video displays. This specification does not specify a pure power curve (at least not for displays with imperfect black), it specifies a custom curve with a well-defined electro-optical transfer function.

Quote:
Originally Posted by James Freeman View Post
Question: Why madshi included BT.709 Display Gamma into MadVR?
That's a good question. Display gamma suffered from much confusion until ITU-R BT.1886 cleared everything up for us. Ideally madshi should remove BT.709 from the gamma options (because it is confusing and doesn't actually make sense) and add BT.1886 support, but that's a feature request, and last time I checked madshi isn't accepting feature requests right now. In the mean time, it is possible to use BT.1886 by generating a 3DLUT with Argyll.
e-t172 is offline   Reply With Quote
Old 28th September 2013, 21:40   #20156  |  Link
cvrkuth
Registered User
 
Join Date: Apr 2011
Posts: 49
I update my OS (Windows 8.1 Pro), I have same problem again.
I noticed that when playing 1080p24 (mkv, m2ts, ts .. full screen exclusive mode, default settings) my screen goes to 23hz. Because of this, there is a drop frames every 30-40 sec.

If I uncheck 'Present Several frames in advance "(old path), my screen stays in 24hz.

Quote:
Originally Posted by madshi View Post
It works for me. Which GPU do you have? Are you using the latest madVR build? If so, create a debug log, zip it and upload it somewhere and I'll have a look.
Log file here.
I have Sapphire Radeon HD 5670 Ultimate 1GB,
Intel Core2 Duo E8400 3.00GHz, RAM 8 Gb
Panasonic TX-P42G20E, Catalyst™ 13.9
Win8.1 Pro x64, JRiver Media Center 18 Ver. 18.0.212, madVR v0.86.11
__________________
Intel Core i7-4790 CPU @ 3.60GHz, RAM 32 GB Dual-Channel DDR3 @ 665MHz (9-9-9-24),
Panasonic TX-P42G20E, NVIDIA GeForce GTX 970, Win 10 Pro x64,
PotPlayer 1.7.16291 64-bit, madVR v0.92.17

Last edited by cvrkuth; 28th September 2013 at 21:56.
cvrkuth is offline   Reply With Quote
Old 28th September 2013, 21:45   #20157  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by James Freeman View Post
Your Display calibration should be a Pure Power Curve of 2.2 to 2.4.
Your display should be calibrate to BT.1886, which adjusts the gamma curve based on the capabilities of your display.

If your display has 10,000:1 native contrast or greater, you should be using a straight 2.4 power curve.

Quote:
Originally Posted by James Freeman View Post
Question: Why madshi included BT.709 Display Gamma into MadVR?
Because people requested it. And it can work reasonably well if people are still using CRTs without external LUT correction. (the 3DLUT support doesn't allow for enough customization for CRTs)
6233638 is offline   Reply With Quote
Old 29th September 2013, 06:59   #20158  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,249
Quote:
Originally Posted by madshi View Post
Here's one new feature:

http://forum.doom9.org/showthread.php?p=1645630#post1645630
One less essential filter active in Avisynth, awesome.

Some questions:
Will it have values that can be adjusted?
Will the deband be done before or after resizing? And have you compared quality of deband before and after resizing?
What's the general performance impact?
Will there be a an HQ mode at all?
ryrynz is online now   Reply With Quote
Old 29th September 2013, 07:00   #20159  |  Link
James Freeman
Registered User
 
Join Date: Sep 2013
Posts: 919
Quote:
Originally Posted by e-t172 View Post
Not quite. Ideally it should follow the REC-BT.1886 specification,
Quote:
Originally Posted by 6233638 View Post
Your display should be calibrate to BT.1886, which adjusts the gamma curve based on the capabilities of your display.

Thank you for clarifing this.


In the Rec. ITU-R BT.1886, Appendix 1,
Its called EOTF-CRT matching.

I calibrate with Argyll+DispCalGUI, I think Black output offset setting which is 100% offset correction for LCDs and 0% for CRT monitors.

EDIT: Apparently to calibrate to the BT.1886 standard, "Black output offset" should be set to 0 with Gamma of 2.4.
I'll try it tonight and report back.
End of Edit.

This corrects higher output black response (around 0.1 cd/m2) of an LCD in relation to the input to get a more accurate calibration.
Meaning: No crushed blacks, simmilar to an CRT curve, and maximum Contrast Ratio that an LCD capable of.


A BT.709 ecncoding curve is closer to 0.51 (1.96).

0.51 * 2.2 = 1.122 Its already darker than a linear representation.
0.51 * 2.4 = 1.224 Even darker.

According to Rec. ITU-R BT.1886,
The recommended gamma should be set around 2.4.
But I find the end-to-end gamma of 1.22 (2.4) is a little too dark.
When watching movies I sense 2.2 is just right (darkened room), sometimes around 2.0 is even better for shdow detail (with the lights On).


Thanks.

Last edited by James Freeman; 29th September 2013 at 10:24.
James Freeman is offline   Reply With Quote
Old 29th September 2013, 07:43   #20160  |  Link
baii
Registered User
 
Join Date: Dec 2011
Posts: 180
Hi, I just tried the CRU recommend few pages ago, and my u2711 finally seem to able to do 23/24 hz

madvr report 23 and 24 as well, just curious, is it possible that the monitor scaler scale the 23/24 hz into 60hz? Or what is reported by madvr or reclock is definitely right?

Sorry for kind of OP.
baii is offline   Reply With Quote
Reply

Tags
direct compute, dithering, error diffusion, madvr, ngu, nnedi3, quality, renderer, scaling, uhd upscaling, upsampling

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

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

Forum Jump


All times are GMT +1. The time now is 07:04.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, vBulletin Solutions Inc.