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 10th January 2016, 01:46   #35061  |  Link
Thunderbolt8
Registered User
 
Join Date: Sep 2006
Posts: 2,197
Ive heard that you can make kodi start mpc-hc & madvr when you select a file to watch.
__________________
Laptop Lenovo Legion 5 17IMH05: i5-10300H, 16 GB Ram, NVIDIA GTX 1650 Ti (+ Intel UHD 630), Windows 10 x64, madVR (x64), MPC-HC (x64), LAV Filter (x64), XySubfilter (x64) (K-lite codec pack)
Thunderbolt8 is offline   Reply With Quote
Old 10th January 2016, 01:46   #35062  |  Link
foozoor
Registered User
 
foozoor's Avatar
 
Join Date: Feb 2012
Posts: 116
Quote:
Originally Posted by Warner306 View Post
The dilemma madVR has created for this user is that I enjoy Kodi as a media frontend but only the defunct DSPlayer is compatible with madVR. In a year or less, I will be forced to choose between madVR and Kodi. It's madVR, of course.
No need to choose anymore with the new Emby Theater.
This is a home theater application like Kodi. The first beta should be released tomorrow.
As you can see it works with madVR!




Last edited by foozoor; 10th January 2016 at 01:59.
foozoor is offline   Reply With Quote
Old 10th January 2016, 03:00   #35063  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
I requested an improvement to sharpen edges and was told to produce some screenshots to illustrate my specific issues.

After taking some screenshots, I think my concerns about sharpen edges producing excessive ringing are overstated, and the limitation of this sharpener has more to do with the thickness of edges it creates and the resulting silhouette. It is an aggressive sharpener and produces an aggressive result.

Source: The Walk (2015)
  • 720p - 10 GB
  • 1080p - 10 GB
madVR profile settings:

720p
  • Chroma: super-xbr100 + AR
  • Image: Jinc3 + AR
  • Image doubling: Off
  • Upscaling refinement: thin edges (0.1), SuperRes (1)
  • Artifact removal - Debanding: Medium/High
  • Image enhancements: Off
  • Dithering: Ordered

1080p
  • Chroma: super-xbr100 + AR
  • Image: Jinc3 + AR
  • Image doubling: Off
  • Upscaling refinement: Off
  • Artifact removal - Debanding: Medium/High
  • Image enhancements: sharpen edges (0.5)
  • Dithering: Error Diffusion 2
The intent was to show a disparity in the appearance of the image from an upscaled image (720p) enhanced with SuperRes to a source not resized (1080p) but enhanced with sharpen edges.

At a distance the two profiles look fairly equivalent in terms of detail:

(720p) SuperRes (1.0):

Full Size

(1080p) sharpen edges (0.5):

Full Size

But, up close, it becomes apparent why sharpen edges appears so abrasive in motion.

Zoomed-in, the image is clearly of a lower resolution in the 720p shot. The impact of upscaling has softened detail. And, while SuperRes has improved things, it is still soft compared to the 1080p profile.

(720p) SuperRes (1.0):

Full Size

At 1080p, beyond the detail advantage, sharpen edges does produce its own ringing comparable to the 720p shot. But the ringing is more pronounced and the lines around the actor's shirt are noticeably thicker. The increase in resolution has made sharpening less natural and more clinical.

(1080p) sharpen edges (0.5):

Full Size

The same can be said for the next series of shots.

Image detail is reasonably equivalent at a distance.

(720p) SuperRes (1.0):

Full Size

(1080p) sharpen edges (0.5):

Full Size

Zoomed-in, the soft nature of the 720p image is again apparent. This softness better hides the impact sharpening than the 1080p image. The 1080p image also suffers from rigid edges that are the hallmark of my criticism with sharpen edges.

Added ringing is seen in both images. Although, I wouldn't consider it excessive.

(720p) SuperRes (1.0):

Full Size

(1080p) sharpen edges (0.5):

Full Size

And the last series.

(720p) SuperRes (1.0):

Full Size

(1080p) sharpen edges (0.5):

Full Size

(720p) SuperRes (1.0):

Full Size

(1080p) sharpen edges (0.5):

Full Size

Conclusion: This comparison has pointed out the obvious: Sharpening is best combined with image upscaling to remain natural and avoid oversharpening. I do think the ringing of sharpen edges could be improved, but it is not a whole lot more distracting than the ringing caused by image upscaling. My contention with sharpen edges comes with its very nature to enhance object edges; sometimes, excessively.

This finding has not dissuaded me from using image sharpening with high-quality 1080p content. However, it has shown the harsh, clinical nature of sharpen edges and AdaptiveSharpen may not be for everyone. I, for one, now use crispen edges for 1080p -> 1080p sharpening.

cripsen edges rings just like sharpen edges, but does a better job of not bringing attention to itself. I think the choice comes down to a preference in the style of sharpening, more than anything.

And it should be noted, sharpen edges, while clinical is superior to AdaptiveSharpen, which displays sharpen edges' faults even more.

(1080p) crispen edges (0.6):

Full Size

(1080p) sharpen edges (0.5):

Full Size

Problem solved? I guess for now...

Last edited by Warner306; 10th January 2016 at 03:28.
Warner306 is offline   Reply With Quote
Old 10th January 2016, 03:03   #35064  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
Quote:
Originally Posted by foozoor View Post
No need to choose anymore with the new Emby Theater.
This is a home theater application like Kodi. The first beta should be released tomorrow.
As you can see it works with madVR!



Yes, I'm aware of this project and will check it out.

I still prefer the depth and feature-set of Kodi. Kodi has superior add-on support (add-ons that are compatible with DSPlayer and madVR), better metadata and more customization. That said, if you are watching content and not browsing your library, any player with built-in support for madVR is "good-enough."
Warner306 is offline   Reply With Quote
Old 10th January 2016, 03:07   #35065  |  Link
Warner306
Registered User
 
Join Date: Dec 2014
Posts: 1,127
Quote:
Originally Posted by Thunderbolt8 View Post
Ive heard that you can make kodi start mpc-hc & madvr when you select a file to watch.
Yes, I've written a set-up guide on how to do this (in my signature). But it is nowhere near as good as having madVR built-in to the player. Start and resume points work properly, the GUI is consistent and the library is easier to manage in terms what is and isn't watched. Also, there is an actual GUI; MPC doesn't really offer anything in terms of a home theater GUI.

Last edited by Warner306; 10th January 2016 at 03:19.
Warner306 is offline   Reply With Quote
Old 10th January 2016, 03:50   #35066  |  Link
Della
Registered User
 
Join Date: Dec 2011
Posts: 32
Quote:
Originally Posted by Warner306 View Post
He is probably experiencing this with SD -> 1080P upscaling. I don't think it looks like a Blu-ray, but the chain of Image Doubling + SuperRes(3) + crispen edges(1.0) looks pretty impressive to me.

The dilemma madVR has created for this user is that I enjoy Kodi as a media frontend but only the defunct DSPlayer is compatible with madVR. In a year or less, I will be forced to choose between madVR and Kodi. It's madVR, of course.
Thanks, I'll check my configuration.

BTW, I've read your guide before, (though I use MBC with MPC-BE vice Kodi) well done.

Last edited by Della; 10th January 2016 at 03:54.
Della is offline   Reply With Quote
Old 10th January 2016, 09:11   #35067  |  Link
RainyDog
Registered User
 
Join Date: May 2009
Posts: 184
Quote:
Originally Posted by Warner306 View Post
The dilemma madVR has created for this user is that I enjoy Kodi as a media frontend but only the defunct DSPlayer is compatible with madVR. In a year or less, I will be forced to choose between madVR and Kodi. It's madVR, of course.
Why won't you still be able to launch MPC/madVR as an external player with Kodi in a year? That's all I've ever done, even when DSPlayer had madVR built in.
RainyDog is offline   Reply With Quote
Old 10th January 2016, 09:23   #35068  |  Link
RainyDog
Registered User
 
Join Date: May 2009
Posts: 184
Quote:
Originally Posted by Warner306 View Post
I requested an improvement to sharpen edges and was told to produce some screenshots to illustrate my specific issues.

After taking some screenshots, I think my concerns about sharpen edges producing excessive ringing are overstated, and the limitation of this sharpener has more to do with the thickness of edges it creates and the resulting silhouette. It is an aggressive sharpener and produces an aggressive result.
Thanks for the interesting comparisons Warner. But you missed a couple of important inclusions in my opinion.

Matching grabs of the untouched source. And sharpen edges in combination with thin edges, especially since you mention yoursef that the main issue with the former is thicker lines which the latter is there to mitigate.
RainyDog is offline   Reply With Quote
Old 10th January 2016, 09:36   #35069  |  Link
RainyDog
Registered User
 
Join Date: May 2009
Posts: 184
On a related note, I always fall back on the sharpness and resolution pattern on the AVS calibration patterns to look at the effects of sharpeners.

Sharpen edges is the first sharpener I've across that actually seems to darken the image slightly. It's instantly noticeable in the dotted boxes in the bottom left and top right when applied. I've never seen another sharpener affect this pattern in way it does actually, it's very manipulative but in a way I've never really seen. Everything's heavily affected from lines to dots to the shapes of edges on text.

In comparison to others, crispen edges adds obvious aliaising and makes text looks jaggy plus brightens the overall image slightly although it's fairly subtle at the default value of 1.0.

Adapative sharpen is the most subtle of the three in terms of altering the image composition. But it also brightens more than the others and also somehow just looks flatter and harsher with actual viewing content.
RainyDog is offline   Reply With Quote
Old 10th January 2016, 19:48   #35070  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by bran View Post
EDIT2: !!! Enabling "Activate SuperRes Filter" is what is yielding the sepia image. !!!
Oh, good to know, will have a look at that later.

Quote:
Originally Posted by GCRaistlin View Post
So let's sum up:
madVR doesn't take into account doubled frame rate when YADIF deinterlacing is enabled.
That's not my fault. I'm switching refresh rates based on the framerate information the decoder provides to me. If that isn't correct, then that's the decoder's fault, not mine.

Quote:
Originally Posted by GCRaistlin View Post
YADIF deinterlacing seems to be better than madVR's.
It's not as simple as that. YADIF is a video mode deinterlacer. Forced film mode in madVR is a pure IVTC algorithm, which only works properly with fully telecined sources. DXVA deinterlacing is supposed to auto detect whether the source is video or telecined film and adjust accordingly.

For telecined sources, usually madVR's forced film mode is the best deinterlacer out of those 3. For video content, usually DXVA looks better than YADIF, but sometimes DXVA stumble and makes mistakes. YADIF doesn't try to figure out which content is film and which is video, so it cannot stumble.

(Furthermore, DXVA deinterlacing seems broken in Crimson drivers atm, as huhn already mentioned.)

Quote:
Originally Posted by mawen1250 View Post
With the latest version (v0.89.19), the matrix of YCgCo video is recognized as BT.709 and results in wrong color.
That seems to be a bug in the decoder. Which one were you using? LAV Video Decoder?

Quote:
Originally Posted by x7007 View Post
It seems the fix is using NVCP Pre renderer 3
I'm not sure what exactly NVCP prerender is, but if it is what I think it is, it should definitely be set to "application controlled" or however that option is named.

Furthermore: Don't use CUVID, use DXVA Copyback instead. And disable GSync, when using madVR.

Quote:
Originally Posted by Plutotype View Post
2.2.3.3.5.2 Dolby Vision
The Dolby Vision video stream is composed of a BDMV HDR video stream and a Dolby Vision enhancement layer video stream. The enhancement layer is an HEVC video stream with embedded Dolby Vision metadata. The Dolby Vision is characterized by the followings:
  • color primaries: BT.2020 with non-constant luminance
  • EOTF(Electro-Optical Transfer Function): SMPTE 2084
  • Bit depth: 12bit (combination of BDMV HDR video stream and Dolby Vision enhancement layer)
  • Enhancement layer video stream: 1920x1080 resolution, same frame rate with the BDMV HDR video stream, 100Mbps or lower together with the BDMV HDR video stream

http://www.blu-raydisc.com/assets/Do...per_150724.pdf
Quote:
Originally Posted by Plutotype View Post
Regarding Dolby Vision, here is a very good info:
http://hdguru.com/dolby-vision-tries...t-bar-for-hdr/
http://www.dolby.com/us/en/technolog...hite-paper.pdf

Will it be technically possible to achieve accurate playback of the UHD BD mastered in Dolby Vision using MPC-HC/LAVfilters/madvr?
Interesting, but these are all whitepapers, which means it's mostly marketing fluff with very little detailed technical information. E.g. the BD-ROM WhitePaper just has one very little paragraph about Dolby Vision, just describing the very basics. We'd need a lot more information to try to implement full Dolby Vision support.

Quote:
Originally Posted by GCRaistlin View Post
madshi, 2 feature requests:
  1. The possibility to choose constant OSD background color. When the video sample has too bright background (like the sky) it's really hard to read OSD from a distance (on the projector).
  2. The delay before madVR switches the display to the appropriate mode. When I try to drop the player's window to the second monitor it often gets dropped back on the primary one because of immediate mode switch.
Both are features that nobody else has requested before. Which makes me think that they're probably not overly important atm. I need to prioritize development, get the most important stuff done first.

Quote:
Originally Posted by Warner306 View Post
Conclusion: This comparison has pointed out the obvious: Sharpening is best combined with image upscaling to remain natural and avoid oversharpening. I do think the ringing of sharpen edges could be improved, but it is not a whole lot more distracting than the ringing caused by image upscaling. My contention with sharpen edges comes with its very nature to enhance object edges; sometimes, excessively.

This finding has not dissuaded me from using image sharpening with high-quality 1080p content. However, it has shown the harsh, clinical nature of sharpen edges and AdaptiveSharpen may not be for everyone. I, for one, now use crispen edges for 1080p -> 1080p sharpening.

cripsen edges rings just like sharpen edges, but does a better job of not bringing attention to itself. I think the choice comes down to a preference in the style of sharpening, more than anything.

And it should be noted, sharpen edges, while clinical is superior to AdaptiveSharpen, which displays sharpen edges' faults even more.

Problem solved? I guess for now...
Thanks for the interesting images. I do miss the original size BMPs/PNGs, because only that way would I be able to reproduce your sharpening results here, and maybe try different settings or try to improve the algorithms with the help of your images.

Quote:
Originally Posted by RainyDog View Post
On a related note, I always fall back on the sharpness and resolution pattern on the AVS calibration patterns to look at the effects of sharpeners.

Sharpen edges is the first sharpener I've across that actually seems to darken the image slightly. It's instantly noticeable in the dotted boxes in the bottom left and top right when applied. I've never seen another sharpener affect this pattern in way it does actually, it's very manipulative but in a way I've never really seen. Everything's heavily affected from lines to dots to the shapes of edges on text.

In comparison to others, crispen edges adds obvious aliaising and makes text looks jaggy plus brightens the overall image slightly although it's fairly subtle at the default value of 1.0.

Adapative sharpen is the most subtle of the three in terms of altering the image composition. But it also brightens more than the others and also somehow just looks flatter and harsher with actual viewing content.
So short summary: You sound like you hate all of those algos?

FWIW, I've noticed myself that "sharpen edges" might be slightly too dark. It will be ever so slightly brighter in the next build. However, I'm not sure if anything will change for artificial test patterns. Due to the way "sharpen edges" works internally, it might not look too great on such test patterns.

I do like artificial test patterns to get an idea of what an algorithm does exactly. But final judgement of an algo's quality should be made with real life content. After all, if you sit down at an evening to use madVR for your entertainment, you're going to watch movies and not look at test patterns. That said, a good test can be movies with hard coded subtitles. They make a good compromise between artificial test patterns and filmed content. They're usually very sharp, but still often have a bit of anti-aliasing in them, which makes them more suitable to sharpening tests than 1 pixel on/off patterns.
madshi is offline   Reply With Quote
Old 10th January 2016, 20:10   #35071  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Pleeease also updated Adaptive Sharpen with next build?
Could you also take a look again at strength/brightness differences with AS in madVR/media player shader?
If not, of course I can test it again.
aufkrawall is offline   Reply With Quote
Old 10th January 2016, 20:14   #35072  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
madVR v0.90.0 released

http://madshi.net/madVR.zip

Code:
* added support for 3D decoders (IMediaSample3D)
* added support for 3D frame packed playback via HDMI 1.4+ (requires win8.1+)
* added support for native 10bit 4:2:0 DXVA decoding (+ scaling)
* added "Reconstruction" chroma upsampling algorithm (replaces NEDI option)
* updated SuperRes algorithm for chroma upscaling
* replaced luma upscaling "linear light" with "sigmoidal light" option
* sharpen edges is now ever so slightly brighter than before
* updated AdaptiveSharpen to version 2016-01-07
* fixed: D3D11 presentation crashed NVidia driver if 3D vision was enabled
* if DCI-P3 3dlut isn't selected, BT.2020 3dlut is used instead (if available)
* if BT.2020 3dlut isn't selected, DCI-P3 3dlut is used instead (if available)


Finally we have 3D playback support for 3D Blu-Rays!!

In addition to madVR v0.90.0 you will also need to use this latest LAV special build:

https://files.1f0.de/lavf/nightly/LA...-0.67.0-54.exe

Please say thank you to nevcairiel for being willing to implement 3D splitting and decoding, even though his personal interest in 3D playback is not very high. FWIW, this LAV build is a test build, not a release build, and it has a few limitations:

1) It currently only supports 3D MKV files created by MakeMKV. Support for 3D m2ts files will be coming soon, though.
2) Decoding is done by using the Intel Media SDK. So an external Intel DLL is needed, which the installer downloads automatically.
3) It's a very first build, so a few bugs may still exist.

One note to NVidia users: If you have 3D Vision enabled, then the NVidia driver will try to convert 2D movies to 3D when using D3D11 for presentation. This actually works, but how good the conversion is I don't really know. Just be aware that you should disable 3D Vision for 2D movies if you want to watch them in 2D.

The new "Reconstruction" chroma upsampling algorithm is somewhat related to "Bilateral" in that it tries to use the luma channel as a guide to improve the chroma channel quality. However, "Reconstruction" should be a bit more reliable than "Bilateral", showing fewer artifacts and not completely falling apart with some clips. On the other hand, if Bilateral works, it's almost unbeatable in its quality.

"Reconstruction" is based on feisty2's "Chroma Reconstructor", especially the "sharp" and "placebo" options. The default "soft" option is heavily modified by me and optimized for better performance and persistance against artifacts. The "sharp" and "placebo" options often produce very good quality, but sometimes they produce artifacts similar to Bilateral, sometimes not as bad as Bilateral, sometimes even worse. So my recommendation is to use the "soft" variant.

Last edited by madshi; 10th January 2016 at 20:26.
madshi is offline   Reply With Quote
Old 10th January 2016, 20:33   #35073  |  Link
DragonQ
Registered User
 
Join Date: Mar 2007
Posts: 934
If I set Chroma to "Reconstruction", the stats say it's using "Jinc". :/
__________________
TV Setup: LG OLED55B7V; Onkyo TX-NR515; ODroid N2+; CoreElec 9.2.7
DragonQ is offline   Reply With Quote
Old 10th January 2016, 20:51   #35074  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by DragonQ View Post
If I set Chroma to "Reconstruction", the stats say it's using "Jinc". :/
That's just a cosmetical error. Will be fixed next weekend. It's really using "Reconstruction".
madshi is offline   Reply With Quote
Old 10th January 2016, 21:09   #35075  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Amazing changelog!
With my sh*tty cartoon samples full of ringing and other artifacts, the new chroma reconstruction looks very good with the soft setting.
To which other algorithm might the performance hit be compareable?
aufkrawall is offline   Reply With Quote
Old 10th January 2016, 21:19   #35076  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,903
there is a new auto scaling OSD feature right?

i don't see it in the changelog.
huhn is offline   Reply With Quote
Old 10th January 2016, 21:20   #35077  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
Quote:
Originally Posted by huhn View Post
there is a new auto scaling OSD feature right?

i don't see it in the changelog.
That was in the previous version already.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 10th January 2016, 21:23   #35078  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
With what bitdepth does the DXVA2 10 bit scaling work?
aufkrawall is offline   Reply With Quote
Old 10th January 2016, 21:33   #35079  |  Link
madshi
Registered Developer
 
Join Date: Sep 2006
Posts: 9,140
Quote:
Originally Posted by aufkrawall View Post
Amazing changelog!
With my sh*tty cartoon samples full of ringing and other artifacts, the new chroma reconstruction looks very good with the soft setting.
To which other algorithm might the performance hit be compareable?
Good to hear you like it! I actually used your favorite cartoon sample as one of the test samples for fine tuning the algorithm.

In terms of performance, I would expect it to be slower than Jinc and super-xbr, but faster than NNEDI3-16.

Quote:
Originally Posted by huhn View Post
there is a new auto scaling OSD feature right?

i don't see it in the changelog.
Quote:
Originally Posted by nevcairiel View Post
That was in the previous version already.
^ True, I guess I should update the changelog for previous version.

Quote:
Originally Posted by aufkrawall View Post
With what bitdepth does the DXVA2 10 bit scaling work?
Well, IIRC for NVidia I'm using P010 input and 16bit RGB output. Which exact bitdepth the DXVA scaler internally uses I don't know. NVidia hasn't had great DXVA scaling, though, from what I remember, so it's probably not a good choice atm, and I suppose that AMD and Intel will only start supporting 10bit scaling once they also support 10bit DXVA HEVC decoding.
madshi is offline   Reply With Quote
Old 10th January 2016, 21:48   #35080  |  Link
aufkrawall
Registered User
 
Join Date: Dec 2011
Posts: 1,812
Quote:
Originally Posted by madshi View Post
Good to hear you like it! I actually used your favorite cartoon sample as one of the test samples for fine tuning the algorithm.
Nice to hear it might have been helpful.

Quote:
Originally Posted by madshi View Post
In terms of performance, I would expect it to be slower than Jinc and super-xbr, but faster than NNEDI3-16.
I've used NNEDI3 32 for some occasions, so I replace it with Reconstruction now (and will report back if I see anything worth mentioning).

The changes to sharpen edges to avoid darkening look good with the FineSharp test image. With the old sharpen edges I also noticed the darkening with game screenshots. Will test this again with the new algorithm.
But first an Adaptive Sharpen comparison madVR vs. MPC HC shader.

Btw: Thanks for updating it.
aufkrawall 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 20:07.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.