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 18th January 2013, 16:57   #17021  |  Link
iSunrise
Registered User
 
Join Date: Dec 2008
Posts: 497
Quote:
Originally Posted by madshi View Post
However, I agree with DragonQ that "proper" video deinterlacing is much better than simple Bob. I don't know of any consumer hardware solution which uses motion compensation for video mode deinterlacing (not sure about what ATI and NVidia do, though, maybe they do some sort of motion compensation, I don't know). The deinterlacer hardware chips in TVs, Receivers, Blu-Ray players etc all use motion-adaptive deinterlacing with some sort of diagonal filtering. The key feature of motion-adaptive deinterlacing is that it detects static vs. moving image areas. For static image areas you can safely weave the fields together without getting artifacts and as a result for static image areas you get full progressive resolution. For moving parts most implementations use a Bob scaling algorithm with diagonal filtering, which produces results which are better than a Bob algorithm based on simple Bilinear scaling.

I would guess that if you used something like Jinc on every separate field (like 6233638 seems to suggest), you'd get better image quality for *moving* parts than what the typical CE deinterlacer chip does. However, if you don't weave static image areas then you're going to lose a lot of resolution during "quiet" scenes.
Interesting. I always wondered why some interlaced material looks overly soft after it got deinterlaced. I originally thought that deinterlacing halves the frame rate but at least puts the frames together as close to the original as possible. I didnīt know that thereīs also filtering applied which explains some of the results Iīm getting.

That diagonal filtering also seems to occur with NV GPUs a lot. At least I usually get a WAY sharper image (more fine details) in the interlaced picture compared to active deinterlacing with a lot of my DVDs (I have about 500-600 DVDs that are NTSC, about 100 are PAL). Also, for some reason, on some DVDs, I donīt need to deinterlace at all and have a perfect and pretty sharp picture.

After years of madVR use, I also noticed that sharp algorithms like Lanczos/Bicubic on luma amplify some of the interlacing artifacts a lot, to the point where it gets overly distracting, while SoftCubic makes some of these artifacts to entirely go away. Which probably is one of the reasons why you included it in the first place, because it can hide artifacts. I really hate itīs desaturation, though, because the picture not only looks very soft afterwards, but also a bit dull. If you have a movie with a lot of bright lights (traffic lights in a big city or dark areas with bright lights, like large city camera shots, which there happen to be a lot when you have a TV series that got produced in the US) you will reduce the dynamics a lot and the picture looks almost "dead" afterwards. I think that once we get to OLEDs, the desaturation effect on SoftCubic is going to get even more pronounced, since OLEDs have a much higher contrast. SoftCubic also looked horrible on my old CRT, because CRTs already had quite a soft picture (compared to an LCD), because of the way they produce the image.

If there were some kind of method we could apply that acts like SoftCubic but without the desaturation effect, SoftCubic would work a lot better. Is there any possibility to counter that desaturation either in the SoftCubic algorithm itself (which is probably a result of the interpolation it does) so that edges that get overly softened and desaturated at least get a boost in saturation again. Is it possible to calculate the amount of desaturation according to the percent on SoftCubic (50%-100%) we choose in the madVR scaling options?

Last edited by iSunrise; 18th January 2013 at 17:07.
iSunrise is offline   Reply With Quote
Old 18th January 2013, 16:58   #17022  |  Link
Qaq
AV heretic
 
Join Date: Nov 2009
Posts: 422
Quote:
Originally Posted by xabregas View Post
(2) yes, guess what, madvr started being smooth after disabling reclock.
You just had to disable vsync in ReClock.
Qaq is offline   Reply With Quote
Old 18th January 2013, 17:53   #17023  |  Link
Dodgexander
Registered User
 
Join Date: Jul 2008
Posts: 157
Quote:
Originally Posted by DragonQ View Post
Is there a particular reason you want to use YADIF? In most cases hardware deinterlacing is better.
I seem to be trying the impossible.

Trying to set things up to so both SD and HD look good. Problem is, the settings in CCC that make SD look good, make HD look odd.

So I thought if I use yadif, it disables the AMD CCC settings so I can leave the HD settings on without losing what I consider to be good SD picture quality.

I would purely use yadif, but my PC can't handle running 1080i with it on.

The other problem I am trying to combat is motion when watching sport at 1080i. At the moment I find watching football in HD is so blurry and slow compared to SD. Ive tried the dimitri filter and also SVP, but they make the ball and players blurry.

I think what I am looking for is the best way to deinterlace SD content compared to HD.
Dodgexander is offline   Reply With Quote
Old 18th January 2013, 18:33   #17024  |  Link
DragonQ
Registered User
 
Join Date: Mar 2007
Posts: 930
Quote:
Originally Posted by iSunrise View Post
That diagonal filtering also seems to occur with NV GPUs a lot. At least I usually get a WAY sharper image (more fine details) in the interlaced picture compared to active deinterlacing with a lot of my DVDs (I have about 500-600 DVDs that are NTSC, about 100 are PAL). Also, for some reason, on some DVDs, I donīt need to deinterlace at all and have a perfect and pretty sharp picture.
Most DVDs are of films and are thus progressive in nature and don't need deinterlacing. They should be IVTCed instead but the current dispute is whether 2:2 cadence on PAL DVDs (i.e. the fields just need weaving together) is detected properly in different situations.

Quote:
Originally Posted by Dodgexander View Post
I seem to be trying the impossible.

Trying to set things up to so both SD and HD look good. Problem is, the settings in CCC that make SD look good, make HD look odd.

So I thought if I use yadif, it disables the AMD CCC settings so I can leave the HD settings on without losing what I consider to be good SD picture quality.

I would purely use yadif, but my PC can't handle running 1080i with it on.

The other problem I am trying to combat is motion when watching sport at 1080i. At the moment I find watching football in HD is so blurry and slow compared to SD. Ive tried the dimitri filter and also SVP, but they make the ball and players blurry.

I think what I am looking for is the best way to deinterlace SD content compared to HD.
Hmm. What settings are you fiddling with in the Catalyst Control Centre? I find most post-processing algorithms (offered by nVidia at least) do more harm than good. I'd just set deinterlacing to "vector adaptive" and leave the rest disabled.

Also, have you checked your CPU and GPU usage during these "blurry and slow" sequences? Could be just lack of processing power.
__________________
HTPC Hardware: Intel Celeron G530; nVidia GT 430
HTPC Software: Windows 7; MediaPortal 1.19.0; Kodi DSPlayer 17.6; LAV Filters (DXVA2); MadVR
TV Setup: LG OLED55B7V; Onkyo TX-NR515; Minix U9-H
DragonQ is offline   Reply With Quote
Old 18th January 2013, 18:34   #17025  |  Link
xabregas
Registered User
 
Join Date: Jun 2011
Posts: 119
Quote:
Originally Posted by Qaq View Post
You just had to disable vsync in ReClock.
It was disabled, somehow ATI HDMI audio + wasapi + reclock = fail

or better said:

ati hdmi audio + wasapi + reclock + ME = FAIL
xabregas is offline   Reply With Quote
Old 18th January 2013, 19:11   #17026  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,137
Quote:
Originally Posted by DragonQ View Post
Is there a particular reason you want to use YADIF? In most cases hardware deinterlacing is better.
Based on what evidence? You may be right, but I fear many people judged based on Cheese Slices and that is just not valid way to judge video deinterlacing quality. Or did you test with real interlaced video content to come to the conclusion that hardware deinterlacing is better than YADIF in most cases?

Quote:
Originally Posted by xabregas View Post
Quote:
Originally Posted by mzso
Don't get to bothered. It often sucks with nvidia too. Sometimes the video gets jerky no matter what renderer you use. Only a system or driver restart helps.
damn
Don't give up too early. What mzso reports is not normal. Most madVR users have no problems whatsoever, from what I can say. Maybe a bit of tweaking is needed sometimes to get things up and running nicely, but that's about it, in most cases (if your PC is fast enough).

Quote:
Originally Posted by xabregas View Post
i have to disable desktop composition in general settings when using exclusive mode with aero enabled or is fine with only exclusive mode checked?
You don't have to disable it. You can try if it works better that way.

Quote:
Originally Posted by xabregas View Post
(1) i always try with software decoding, tried with lav and coreavc, coreavc is much more smooth
That's weird, is your CPU old and slow? There should be no big difference between LAV and CoreAVC, unless your CPU is very old, then maybe CoreAVC is ever so slighty faster, but not much (if at all).

Quote:
Originally Posted by xabregas View Post
(2) yes, guess what, madvr started being smooth after disabling reclock. sound is bad now, but i went to the loft and brought my old pci xfi xtrememusic and sound is good again, no need for reclock
Then I would guess that something is wrong with the Reclock configuration. You don't necessarily have to keep it off. Maybe clearing the timing database and playing with the Reclock settings could fix the issue? But I can't help there much, my Reclock experience is rather limited.

Quote:
Originally Posted by iSunrise View Post
Interesting. I always wondered why some interlaced material looks overly soft after it got deinterlaced. I originally thought that deinterlacing halves the frame rate but at least puts the frames together as close to the original as possible. I didnīt know that thereīs also filtering applied which explains some of the results Iīm getting.

That diagonal filtering also seems to occur with NV GPUs a lot. At least I usually get a WAY sharper image (more fine details) in the interlaced picture compared to active deinterlacing with a lot of my DVDs (I have about 500-600 DVDs that are NTSC, about 100 are PAL). Also, for some reason, on some DVDs, I donīt need to deinterlace at all and have a perfect and pretty sharp picture.
The diagonal filtering I mentioned does not reduce sharpness, it just removes jagging artifacts in moving parts of the image. The diagonal filtering is all good, not bad. It's also only used for video mode deinterlacing, not for movie content (unless your GPU failed to detect that the source is a movie).

Quote:
Originally Posted by iSunrise View Post
Is it possible to calculate the amount of desaturation according to the percent on SoftCubic (50%-100%) we choose in the madVR scaling options?
That won't work. SoftCubic does not desaturate by itself, it just has some blur included. Which means that small colored highlights get desaturated (because of the blur). But bigger colored areas do *not* get desaturated at all. So you can't just up the saturation control to fix things. Doing so would fix small colored highlights. But then big colored areas (e.g. sky, grass, water etc) would get too much saturation.
madshi is offline   Reply With Quote
Old 18th January 2013, 19:21   #17027  |  Link
Dodgexander
Registered User
 
Join Date: Jul 2008
Posts: 157
Turns out I was comparing the settings with "dynamic contrast" on in CCC and I didn't like the difference it was making between Yadif and Dxva deinterlacing.

Ill stick with hardware deinterlacing for now then, unless there is some software alternatives of better quality I can try. I did like ffdshows ffmpeg deinterlacer that worked well for my football 1080i sample.

Hey fiddlings the game, HTPCs the name
Dodgexander is offline   Reply With Quote
Old 18th January 2013, 19:40   #17028  |  Link
DragonQ
Registered User
 
Join Date: Mar 2007
Posts: 930
Quote:
Originally Posted by madshi View Post
Based on what evidence? You may be right, but I fear many people judged based on Cheese Slices and that is just not valid way to judge video deinterlacing quality. Or did you test with real interlaced video content to come to the conclusion that hardware deinterlacing is better than YADIF in most cases?
Well I tested this a long time ago myself but that's only for one family of GPUs. AMD could well be different. Intuitively it makes sense that hardware deinterlacing would be better but I guess more tests need to be done.

Maybe we should get some clips and get people to upload screenshots with specific settings?
__________________
HTPC Hardware: Intel Celeron G530; nVidia GT 430
HTPC Software: Windows 7; MediaPortal 1.19.0; Kodi DSPlayer 17.6; LAV Filters (DXVA2); MadVR
TV Setup: LG OLED55B7V; Onkyo TX-NR515; Minix U9-H
DragonQ is offline   Reply With Quote
Old 18th January 2013, 19:48   #17029  |  Link
Dodgexander
Registered User
 
Join Date: Jul 2008
Posts: 157
Quote:
Originally Posted by madshi View Post
The cause of the crash is that your filter does implement "IFileSourceFilter" and "IFileSourceFilter.GetCurFile()" returns success, but the returned file name is NULL. madVR then crashes when later calling CoTaskMemFree() to properly release the file name. You can fix the crash by either returning a proper file name, or by letting GetCurFile() return an error, or by not supporting IFileSourceFilter. I'll add a check for a NULL file name to the next build. But I think you should implement the suggested change in your filter, anyway.
I saw this, I use NPVR and for some reason, although it seems to work. The OSD of NPVR doesn't dissapear.

Also for some reaons with Madvr as the render, if you launch in fullscreem mode it doesn't scale and just shows a window of the image, until you resize and make fullscreen again.

Any ideas??
Dodgexander is offline   Reply With Quote
Old 18th January 2013, 19:56   #17030  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,137
Quote:
Originally Posted by DragonQ View Post
Well I tested this a long time ago myself but that's only for one family of GPUs. AMD could well be different. Intuitively it makes sense that hardware deinterlacing would be better but I guess more tests need to be done.

Maybe we should get some clips and get people to upload screenshots with specific settings?
I think it would be interesting. I'd really like to know how YADIF compares to the best AMD and NVidia can do. But maybe a new thread should be started for this, because it's not really madVR related. In any case, the first step would be to collect samples which show clear differences between YADIF and DXVA.

BTW, for NVidia and ATI, hardware deinterlacing is really using the pixel shaders, not hard wired circuits. For Intel I think there's a dedicated circuit for deinterlacing, but what I've been told it's not as good as NVidia's and ATI's deinterlacing. JFWIW.

Quote:
Originally Posted by Dodgexander View Post
I saw this, I use NPVR and for some reason, although it seems to work. The OSD of NPVR doesn't dissapear.

Also for some reaons with Madvr as the render, if you launch in fullscreem mode it doesn't scale and just shows a window of the image, until you resize and make fullscreen again.

Any ideas??
You should ask the NPVR developer about this, not me.
madshi is offline   Reply With Quote
Old 18th January 2013, 21:58   #17031  |  Link
leeperry
Kid for Today
 
Join Date: Aug 2004
Posts: 3,463
It's very convenient to have mVR automatically hiding the mouse cursor after a few secs in FSE mode, but do you really find it to be mandatory for windowed playback? I mean you could let the movie play as "always on top" in the center of the screen in a small window and do other stuff simultaneously, atm it's a major PITA as the mouse cursor keeps disappearing

And let's not mention ppl with dual screens setups.

Last edited by leeperry; 18th January 2013 at 22:06.
leeperry is offline   Reply With Quote
Old 18th January 2013, 22:05   #17032  |  Link
ThurstonX
Registered User
 
Join Date: Mar 2006
Posts: 58
Quote:
Originally Posted by madshi View Post
Please disable "forcing smooth playback". Nobody even knows what that does exactly.
Quote:
Originally Posted by xabregas View Post
Ok, will disable it
I'm not saying this is the definitive answer on ATI's/AMD's "Enforce Smooth Video Playback (ESVP)" feature, but it's an interesting read which includes feedback from AMD:

http://www.anandtech.com/show/3601/t...deon-htpc-card

"The short answer is that it’s an “idiot button” for users, to keep them from using so many post-processing settings on low-end video cards that It causes frames to drop."

There's more detail at that link.
ThurstonX is offline   Reply With Quote
Old 18th January 2013, 23:45   #17033  |  Link
Dodgexander
Registered User
 
Join Date: Jul 2008
Posts: 157
Using Mad vr with Media Player Classic and FFDshow DXVA video decoder results in a crash of mpc when moving to a second screen, or setting media player classic to start playback on a secondry screen.
Dodgexander is offline   Reply With Quote
Old 19th January 2013, 00:20   #17034  |  Link
Matching_Mole
Registered User
 
Join Date: Feb 2009
Posts: 99
I coming back regarding my "lack of smoothness" of the IVTC due to my higher screen refresh frequency.

Indeed, I tried several player+video decoding+splitter but,as some one else has already mentioned, no combination solved the issue. I tried also to use Reclock "speed adaption" feature with the same result.

I have a question regarding Reclock and its use with IVTC. I mean when I used it with a progressive film, the clock deviation reported by madVr is reduced dramatically to almost 0, which is expected. I payed also attention to the clock deviation with a telecined film and with or without Reclock it was around -0.4% so I was not sure of the effect of Reclock. Though Reclock indicated that it try to adapt the speed to 28.8fps instead of 29.97.

Someone can help me to understand if Reclock see the fps change due to the madVr IVTC process or if use Reclock with madVr IVTC process is more hurting than help? And with the IVTC process used, what should be the correct clock deviation expected with a refresh rate of 48hz?
Matching_Mole is offline   Reply With Quote
Old 19th January 2013, 01:36   #17035  |  Link
xabregas
Registered User
 
Join Date: Jun 2011
Posts: 119
Quote:
Originally Posted by ThurstonX View Post
I'm not saying this is the definitive answer on ATI's/AMD's "Enforce Smooth Video Playback (ESVP)" feature, but it's an interesting read which includes feedback from AMD:

http://www.anandtech.com/show/3601/t...deon-htpc-card

"The short answer is that it’s an “idiot button” for users, to keep them from using so many post-processing settings on low-end video cards that It causes frames to drop."

There's more detail at that link.
nice catch

now i know i can disable it because i disabled every default post processing on CCC

And do anyone tried amd steady video??

I checked once and with dxva decoding i saw the amd steady logo when video started, looked very smooth, so much smooth the audio went OOS after 5 min and after 10 the movie dropped idk how many frames

Anyway is interesting to see how amd techs find some new settings to help us screwing video playback

Fingers crossed now has i`m going to unninstall amd drivers (FEAR and panic), and then install 13.1 brand new drivers, so happy i am, maybe they have a surprise in this drivers for HD7770 owners, idk, some new amazing setting to delete movies after playback

Last edited by xabregas; 19th January 2013 at 01:39.
xabregas is offline   Reply With Quote
Old 19th January 2013, 03:17   #17036  |  Link
Omega13Lives
Registered User
 
Join Date: Mar 2003
Posts: 2
Spanning multiple monitors

Don't know if this is where to post this. The folks at ZoomPlayer suggested I contact the developer of madVR for my question. Here:

There is a cool program called SoftTH which allows you to use Direct3D to span multiple monitors for video games AND you can specify the bezel/border size so the picture looks consistent like you were watching (at a drive-in, if anyone knows what that is) through a car windshield with the frame blocking part of the view but the picture/video looks normal when you move or turn. With ZP Video Wall I can stretch over 3 monitors but the video gaps to the outside monitors. Is it possible to adjust for the gaps/bezels to span multiple monitors AND allow for the bezel? tx

I know this is not common, but thought I'd ask

Eric
__________________
I hate computers (when they don't do what I want).
Omega13Lives is offline   Reply With Quote
Old 19th January 2013, 04:57   #17037  |  Link
chuuey
Registered User
 
chuuey's Avatar
 
Join Date: Oct 2007
Posts: 87
Ineteresting, I've tried using madvr on the same machine but in windows xp (i need it for work) and madvr won't even display video with 1080p content, 720p stutters like crazy lol, using lav filters, geforce 670, 6 core cpu... weird... standard renderers play with no issues o.o
chuuey is offline   Reply With Quote
Old 19th January 2013, 09:16   #17038  |  Link
dansrfe
Registered User
 
Join Date: Jan 2009
Posts: 1,212
I know this is something minor but would it be possible to have a mouse pointer cursor (with the pointing finger) in exclusive mode? It seems a bit more natural that way. Possibly something to add to the v1.0 list.

Thanks!
dansrfe is offline   Reply With Quote
Old 19th January 2013, 10:41   #17039  |  Link
6233638
Registered User
 
Join Date: Apr 2009
Posts: 1,019
Quote:
Originally Posted by dansrfe View Post
I know this is something minor but would it be possible to have a mouse pointer cursor (with the pointing finger) in exclusive mode? It seems a bit more natural that way. Possibly something to add to the v1.0 list.
What player are you using? You should still have a mouse cursor.
6233638 is offline   Reply With Quote
Old 19th January 2013, 11:04   #17040  |  Link
secvensor
Registered User
 
Join Date: Sep 2011
Posts: 72
What video card is necessary from the last that it was possible to reproduce 1080p@60fps@10bit@100MHz at jinc 8 taps hroma upsc?
ati ore nv?
7850-7970 ore 660-680
__________________
Win7x64
Core i7 920 3.5GHz
Noctua NH-D14/ArcticCooling MX-3
6(3x2)GB Transcend 1426MHz
RoyalHD (64MB)[Solo6c][JPLAY]/HD7850DC22GD5V2[EIZOT965]
Seasonic X-750
VelociRaptor WD4500HLHX/16TB_STORE

Last edited by secvensor; 19th January 2013 at 11:08.
secvensor 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 04:44.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.