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
Register FAQ Calendar Today's Posts Search

Reply
 
Thread Tools Search this Thread Display Modes
Old 18th January 2016, 18:09   #2501  |  Link
Toco
Registered User
 
Toco's Avatar
 
Join Date: Jan 2014
Posts: 10
According to this commit , the latest nightly should have new decoder based on Intel MediaSDK. Where can I find this new decoder and why may I want to use it? I use Intel Core i5 4200U(HD Graphics 4400) and Intel Atom Z3736F, if it is important.
Toco is offline   Reply With Quote
Old 18th January 2016, 18:13   #2502  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,926
it's for MVC 3d decoding. and not sure if mpc-hc can be shipped with it.

just read the commit.
huhn is offline   Reply With Quote
Old 19th January 2016, 14:34   #2503  |  Link
Matching_Mole
Registered User
 
Join Date: Feb 2009
Posts: 99
CTR+L and xySubfilter

I recently updated my version of MPCHC and seen the better integration of xySubfilter as subtitle renderer. This is a great addition! My only issue is related to the "Load Subtitle CTR+L" function that does not work with xySubfilter (and maybe standard version of vsfilter).

Indeed, I used this function a lot to add on the fly external subtitles to my physical media (Bluray, DVD,...). My expectation is to have the CTR+L function linked to the "open" function of xySubfilter. Does this "link" is already planned or could be added?

Furthermore, I can see that xySubfilter is not loaded when no subtitle is detected at the creation of the graph, meaning that the loading of an external subtitle could be problematic afterward. The two options seems:
- to always load xySubfilter. And the option is already defined in xySubfilter.
- to create a fallback to the internal subtitle renderer but this need to be developped.
What is your view?

Anyway, thank you very much for the huge amount of work on MPCHC!
Matching_Mole is offline   Reply With Quote
Old 20th January 2016, 22:54   #2504  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
I got a question about Full/Half Floating Point Processing.
Is it supposed to make Chroma edges sharper/blockier?

Here is with Full Floating Point Processing On:
http://abload.de/img/haloce_2016_01_20_22_dijjg.png
Here is with it Off:
http://abload.de/img/haloce_2016_01_20_22_33kwp.png

Look at the Red thingy in the middle, it's easiest to detect the difference on it's edges.

Last edited by zerowalker; 24th January 2016 at 14:04. Reason: Removed the Images and just post it's links cause they are too big and i don't know how to resize.
zerowalker is offline   Reply With Quote
Old 22nd January 2016, 08:59   #2505  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,784
There seems to be a different kind of interpolation of the chrominance, which has most often a 4:2:0 subsampling, therefore a reduced resolution in relation to the chrominance, which appears quite obviously for diagonals of very saturated colors, especially red and blue. Modern codecs are optimized for natural content, not for PC games; recordings of natural scenes rarely have so saturated colors with so sharp edges.

But where do you find this "full/half floating point" option? Is it in MPC-HC dialogs, or rather in dialogs of the decoder or renderer (and thus a bit off-topic here)?
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 22nd January 2016, 09:06   #2506  |  Link
huhn
Registered User
 
Join Date: Oct 2012
Posts: 7,926
they are under "renderer settings" in mpc-hc so they are a part of mpc-hc.
huhn is offline   Reply With Quote
Old 23rd January 2016, 15:55   #2507  |  Link
vivan
/人 ◕ ‿‿ ◕ 人\
 
Join Date: May 2011
Location: Russia
Posts: 643
pls don't embed huge images.

That's obviously nearest neighbor upscaled chroma. It might be that your gpu doesn't support bilinear scaling with full precision textures (not sure about desktop ones, but it's a thing in mobile world). Either that, or something just breaks (which is not suprising for EVR).
vivan is offline   Reply With Quote
Old 24th January 2016, 13:57   #2508  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
Thanks for the answers.
Hmm, about my GPU not supporting it, that kinda sucks, feels weird as it's fairly new (R9 380):S
Would be nice if someone else could test it for me if they have time.

How can i make the image smaller btw, cause i didn't really want it to look that big as i know it's a bother.

EDIT: Changed the images to just links.

Btw, what does full precision really do, i get that it works in 32bit or something, but what does it normally work at?

Last edited by zerowalker; 24th January 2016 at 14:06.
zerowalker is offline   Reply With Quote
Old 24th January 2016, 14:42   #2509  |  Link
vivan
/人 ◕ ‿‿ ◕ 人\
 
Join Date: May 2011
Location: Russia
Posts: 643
Quote:
Originally Posted by zerowalker View Post
Thanks for the answers.
Hmm, about my GPU not supporting it, that kinda sucks, feels weird as it's fairly new (R9 380):S
Nah, I was thinking about something ancient, not modern gpus (AFAIK DX10 requires full 32-bit floating point support). So the problem is that something breaks.

Quote:
Originally Posted by zerowalker View Post
Btw, what does full precision really do, i get that it works in 32bit or something, but what does it normally work at?
The other floating point precision is half-float. It uses 16 bits and has at least 11 bits of precision in [0..1] range.
Another option is 8/10/16/32-bit (normalized, means maps to [0..1]) integer textures.
For example madVR uses 16-bit integer textures, but has 10-bit option (in trade quality for performance).

Rendering works this way: input -> processing -> texture -> processing -> texture ... -> output.
There's processing precision (normally full floating point precision, running it at half precision doesn't give any advantage on desktop gpus and should not affect compatibility) and texture (that hold intermediate result) precision (less precision = less bandwidth = faster).

Less precision at any step gives either banding or noise (if dithering is used). How much is enough and how much is more than enough? Who knows, but you can try 10-bit precision in madVR.

Last edited by vivan; 24th January 2016 at 14:45.
vivan is offline   Reply With Quote
Old 24th January 2016, 17:22   #2510  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
Quote:
Originally Posted by vivan View Post
Nah, I was thinking about something ancient, not modern gpus (AFAIK DX10 requires full 32-bit floating point support). So the problem is that something breaks.

The other floating point precision is half-float. It uses 16 bits and has at least 11 bits of precision in [0..1] range.
Another option is 8/10/16/32-bit (normalized, means maps to [0..1]) integer textures.
For example madVR uses 16-bit integer textures, but has 10-bit option (in trade quality for performance).

Rendering works this way: input -> processing -> texture -> processing -> texture ... -> output.
There's processing precision (normally full floating point precision, running it at half precision doesn't give any advantage on desktop gpus and should not affect compatibility) and texture (that hold intermediate result) precision (less precision = less bandwidth = faster).

Less precision at any step gives either banding or noise (if dithering is used). How much is enough and how much is more than enough? Who knows, but you can try 10-bit precision in madVR.
Oh wait, this is confusing.

I mean, if i have 8bit content on an 8bit screen, and i don't do any post-processing etc.
Shouldn't it be displaying lossless by default?
Why would any kind of precision play a role, there shouldn't be any processing right, or am i missing something here?

Cause with 10bit content i understand there being dithering and the like.

But for the nearest-neighbor chroma thingy.
Why would it even do that is the chroma isn't upscaled/downscaled?
I mean, is it's original resolution, shouldn't it be simply what the content has with no processing?

I can btw say that MadVR looks like EVR without full processing,
not taking a close look, it might look better.
Though it still shouldn't right, if it's no post-processing?

This is confusing!:P
zerowalker is offline   Reply With Quote
Old 24th January 2016, 17:56   #2511  |  Link
Shiandow
Registered User
 
Join Date: Dec 2013
Posts: 753
Quote:
Originally Posted by zerowalker View Post
I mean, if i have 8bit content on an 8bit screen, and i don't do any post-processing etc.
Shouldn't it be displaying lossless by default?
Why would any kind of precision play a role, there shouldn't be any processing right, or am i missing something here?
Ideally that would be the case, but even though the content is in 8 bit, it's not in RGB yet, it's in YUV and there's no one-to-one conversion from one to the other. To get an accurate result you'll need to deband (to increase the number of bits), then convert to RGB, and then use dithering (to lower the number of bits again).

Even if it was in RGB it might still be better to deband anyway, to recover the information that was lost when it was converted to 8 bits. And if you want to do any kind of processing (scaling etc.) you'll also need some additional precision to prevent rounding errors, which can actually be quite big even with 16 bit floats.
Shiandow is offline   Reply With Quote
Old 24th January 2016, 19:40   #2512  |  Link
vivan
/人 ◕ ‿‿ ◕ 人\
 
Join Date: May 2011
Location: Russia
Posts: 643
Quote:
Originally Posted by zerowalker View Post
I mean, if i have 8bit content on an 8bit screen, and i don't do any post-processing etc.
Shouldn't it be displaying lossless by default?
Why would any kind of precision play a role, there shouldn't be any processing right, or am i missing something here?
Color conversion (YCbCr -> RGB), chroma upsampling (if video is not 4:4:4), scaling.
There should be no processing only if your video is RGB and there's no scaling.

Quote:
Originally Posted by zerowalker View Post
But for the nearest-neighbor chroma thingy.
Why would it even do that is the chroma isn't upscaled/downscaled?
I mean, is it's original resolution, shouldn't it be simply what the content has with no processing?
Now that you mention it your picture does look better than 4:2:0 (either 4:2:2 or 4:4:4, can't decide).
Can you say what colorspace and subsampling your video has? And also can you check what OSD says (Ctrl+J)?

I've tried EVR again and it forces decoder to output YUY2 (8-bit 4:2:2), since it refuses to accept 4:4:4. And with those half/full-fp precision it just gives no picture...

Last edited by vivan; 24th January 2016 at 20:09.
vivan is offline   Reply With Quote
Old 24th January 2016, 20:08   #2513  |  Link
foozoor
Registered User
 
foozoor's Avatar
 
Join Date: Feb 2012
Posts: 116
Does anyone know how to use VapourSynth with MPC-HC in realtime?

I just get an old core i7 with a miserable iGPU that can't handle madVR.
But VapourSynth uses almost only CPU and it could be used as a kind of madVR alternative for upscaling and image enhancements.
foozoor is offline   Reply With Quote
Old 24th January 2016, 23:53   #2514  |  Link
ryrynz
Registered User
 
ryrynz's Avatar
 
Join Date: Mar 2009
Posts: 3,650
Quote:
Originally Posted by foozoor View Post
Does anyone know how to use VapourSynth with MPC-HC in realtime?

I just get an old core i7 with a miserable iGPU that can't handle madVR.
But VapourSynth uses almost only CPU and it could be used as a kind of madVR alternative for upscaling and image enhancements.
I believe this still cannot be done right now. Use Avisynth or one of its derivatives. MadVR is likely to add support for Vapoursynth in the near future.
ryrynz is offline   Reply With Quote
Old 25th January 2016, 04:42   #2515  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
Quote:
Originally Posted by vivan View Post
Color conversion (YCbCr -> RGB), chroma upsampling (if video is not 4:4:4), scaling.
There should be no processing only if your video is RGB and there's no scaling.

Now that you mention it your picture does look better than 4:2:0 (either 4:2:2 or 4:4:4, can't decide).
Can you say what colorspace and subsampling your video has? And also can you check what OSD says (Ctrl+J)?

I've tried EVR again and it forces decoder to output YUY2 (8-bit 4:2:2), since it refuses to accept 4:4:4. And with those half/full-fp precision it just gives no picture...
Ah okay didn't think of the colorspace.

The video should be 4:2:0 actually.
OSD says UYVY.
zerowalker is offline   Reply With Quote
Old 25th January 2016, 06:34   #2516  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,784
UYVY is like YUY2, just a different order; it has a chroma subsampling of 4:2:2.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 25th January 2016, 19:40   #2517  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
Quote:
Originally Posted by LigH View Post
UYVY is like YUY2, just a different order; it has a chroma subsampling of 4:2:2.
Weird, the file should be YV12 :s

Last edited by zerowalker; 25th January 2016 at 19:56.
zerowalker is offline   Reply With Quote
Old 25th January 2016, 22:15   #2518  |  Link
vivan
/人 ◕ ‿‿ ◕ 人\
 
Join Date: May 2011
Location: Russia
Posts: 643
Quote:
Originally Posted by zerowalker View Post
Weird, the file should be YV12 :s
Not really, I've actually never seen EVR use anything other than YUY2. Though I've heard something about AYUV (4:4:4).
On the one hand it makes it less worse (half of the upsampling is done by LAV on CPU, which is, at least, not broken), on the other hand it's just terrible.
vivan is offline   Reply With Quote
Old 26th January 2016, 00:39   #2519  |  Link
zerowalker
Registered User
 
Join Date: Jul 2011
Posts: 1,121
Quote:
Originally Posted by vivan View Post
Not really, I've actually never seen EVR use anything other than YUY2. Though I've heard something about AYUV (4:4:4).
On the one hand it makes it less worse (half of the upsampling is done by LAV on CPU, which is, at least, not broken), on the other hand it's just terrible.
Oh, this is actually without LAV.
I think when LAV is used i get NV12.
zerowalker is offline   Reply With Quote
Old 27th January 2016, 07:53   #2520  |  Link
EncodedMango
Registered User
 
Join Date: Jun 2013
Posts: 65
I'm having an issue with MPC-HC, from time to time to seems to 'forget' I have chosen that the player window Always stay On Top of everything else. It gets hidden behind other windows and when I check the options it still shows as "Always" selected, but I have to reselect it before it starts to work again as it should. No other settings are affected.

Win10 x64.
EncodedMango is offline   Reply With Quote
Reply

Tags
dxva, media player classic, mpc, mpc-hc


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 23:12.


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