Media Player Classic Home Cinema (MPC-HC) - DXVA!

[sansnom05]

this obscene change should have a configuration option to turn it off!
without easy mouse dragging MPCHC is ****.

This commit is direct answer to this ticket at trac: https://sourceforge.net/apps/trac/mpc-hc/ticket/109 , guess can't make all people happy.

You can still drag mpchc around, just not by click in video window, but by click & drag at different places of bars. And if you use minimal mode, means no bar at all, you can still click the video to drag it.

[pdanpdan]

Is there any way to force the internal decoder to output NV12 to EVR Custom? Now, if I use the internal decoders and DXVA, everything is OK, else I have a black screen. If I use ffdshow and force NV12 it works. ATI HD2400, Catalyst 10.12/11.1a(both tested), Win7 64. It doesn't depend on any option in renderer settings/presentation.

[wOxxOm]

sansnom05,
1. I (personally) can compile my own version (or hack binaries for example)
2. Changes of BASIC behaviour should be made with caution - in this case, MPCHC should have a configuration option.

[sansnom05]

sansnom05,
1. I (personally) can compile my own version (or hack binaries for example)
2. Changes of BASIC behaviour should be made with caution - in this case, MPCHC should have a configuration option.

Since you're an advanced user who can build custom build, this change is only little inconvenience for you. I hope it means you don't really think this change is a blocker then.

About the BASIC behaviour, it depends on who you're asking. For user open that ticket to complain, they think single click means pause/resume playback, and it makes sense, because in "Options->Player->Keys", the default settings is "Play/pause"-> Space key -> Left Down. So another way to solve this is to assign different mouse button to "Play/pause", like Mercury_22 suggested using middle button for that.

:thanks:

[wOxxOm]

About the BASIC behaviour...I mean a behaviour that existed for a few years or more, and suddenly you change it without providing a way back for the majority of users who are accustomed to dragging the window by clicking anywhere.

IMO it makes no sense without providing a configuration option.

[MokrySedeS]

I have to agree with wOxxOm. MPC-HC feels "crippled" for me now... :-/

[sansnom05]

1. you have your configuration option, that is "Options->Player->Keys".

2. you can still drag window in any mode. in fact, my change make no difference for normal/minimal mode, for other modes there are also ways to drag it, just not anywhere, and now it can also play/pause, according to some other users request.

3. play/pause assigned to left mouse, this default settings/behaviour maybe exist longer than borderless/frameonly was introduced IIRC.

[bobdynlan]

This commit is direct answer to this ticket at trac: https://sourceforge.net/apps/trac/mpc-hc/ticket/109 , guess can't make all people happy.

You can still drag mpchc around, just not by click in video window, but by click & drag at different places of bars. And if you use minimal mode, means no bar at all, you can still click the video to drag it.

sansnom05,
2. Changes of BASIC behaviour should be made with caution - in this case, MPCHC should have a configuration option.

What if this BASIC behaviour is flawed?
Play/Pause should be assigned to Left DblClk
- enabling drag everywhere, bars visible or not
- one annoyance less, as you will not Play/Pause every time you open the pop-up menu
I mean a behaviour that existed for a few years or more, and suddenly you change it without providing a way back for the majority of users who are accustomed to dragging the window by clicking anywhere.
- if a monkey was taught to Left Clk to Play/Pause, it will eventually get it, because it will get frustrated and will start to rapidly click the button...
Full-screen should be assigned to Middle Down
- almost every mice has a 3rd button, while touch-pads now days have gestures to simulate it like two-fingers press

And then there are the retarded-feeling default keys for seeking vs. frame-stepping (swap them?) or the audio delay using one of the largest sized key (add Ctrl, or better swap with Pns Inc/Dec Size). Most of these were requested countless times, but will never be applied because of people screaming at every little change.

[wOxxOm]

madness galore... I give up (that is I'll patch it myself).

[bobdynlan]

wOxxOm, don't sweat it, as the specific issue you have (loosed ability that cannot be fixed with key assignement), will surely be reverted :)
My monkey quote was just to point out that anything can change, and anybody can adapt to the changes, but don't want to.

[Mercury_22]

Situation it's simple the left click for Play/pause in compact mode should be fixed but rev 2866 it's NOT the proper fix so I propose to be reverted until a proper fix it's available
And (again) taking in consideration that the left click for pause/play it's very inconvenient (it's staying in the way for drag , enter full screen and so on... ) I (again) propose that the default key to be changed (e.g. to middle down)
AFAIK no other player uses left click for Play / Pause so the new / common users are NOT used that when they click on the image to pause the video and they found this very annoying, (at least those I introduced to the MPC-HC)

Like I've said before breaking one thing to fix another it's not the way to go

[sansnom05]

Situation it's simple the left click for Play/pause in compact mode should be fixed but rev 2866 it's NOT the proper fix so I propose to be reverted until a proper fix it's available
And (again) taking in consideration that the left click for pause/play it's very inconvenient (it's staying in the way for drag , enter full screen and so on... ) I (again) propose that the default key to be changed (e.g. to middle down)
AFAIK no other player uses left click for Play / Pause so the new / common users are NOT used that when they click on the image to pause the video and they found this very annoying, (at least those I introduced to the MPC-HC)

I agree about your default key(for play/pause) should be changed from left click, in fact I always changed them to middle click like you proposed. But if I go ahead change that setting, other peoples will complain I'm sure. Old habits die hard, like wOxxOm or the user who open that tickets show.

I suggest we don't waste more pages on this trivial matters, and I welcome any suggestions to my PM without polluting this thread anymore. :thanks:

[Mercury_22]

I agree about your default key(for play/pause) should be changed from left click, in fact I always changed them to middle click like you proposed. But if I go ahead change that setting, other peoples will complain I'm sure. Old habits die hard, like wOxxOm or the user who open that tickets show.

I suggest we don't waste more pages on this trivial matters, and I welcome any suggestions to my PM without polluting this thread anymore. :thanks:

OK !
Just for the last time :
the most / many of the current users are tech-savvy so it will be easy for them to change the key for Play/pause if they want BUT for the new users it is more difficult and they expect a "normal" usage of the left click out of the box (again, AFAIK no other player uses left click for Play / Pause) and like I've said they are not used with the left click to pause the video and from my experience with them believe or not the annoyance of this is a deal breaker in using MPC-HC !

P.S. And I think this should be considered more from the perspective of a new user than a current user cause after changing the code if you don't tell people to reset their settings they will continue using their current settings so there will be no complaints (until they reset their settings :) )
And now I'll shut up about this ! :)

[Polcius]

In DispcalGUI, I had to install the profile "system-wide" to see any change in MPC-HC with CMS enabled.

If I installed the profile "for only this user", CMS enabled in MPC-HC didn't seem to change anything.

Now the blue primary seems really corrected; I assume its right.

Here's my profile attached.

[janos666]

@Polcius

If you post some measurement data here (http://forum.doom9.org/showthread.php?t=157722), I will try to draw some charts for you (2D CIE gamut view and TRC curves).
I need a spotread log file with XYZ values from uncorrected and corrected states measured in the MPC-HC window. You can use the AVS-HD test videos (http://www.avsforum.com/avs-vb/showthread.php?t=948496) to measure 10 gray-scale patches and the 3 primaries. (You should also clear your VGA LUT for the "uncorrected" measurements.)
Example script:
C:\Argyll\bin\spotread.exe -d -yl -db -H uncorrected.log

[WonderCsabo]

hi!

MPC-HC created screenshots (ALT+I) gone mad. http://itcafe.hu/dl/upc/2011-01/60463_screenshot.png

Using rev 2860 build. Is this a common issue, or am i doing something wrong?

[JanWillem32]

@WonderCsabo The export function tries to put 32-, 16- or 10-bit images into an 8-bit image. It will work if you make a screenshot of the backbuffer (Alt+Print Screen or Print Screen and paste it to an image editor). You can also disable the 32-, 16- or 10-bit surfaces and then save the image. I'm already working on a surface export function that will grab the full contents of the last screenspace surface.

[WonderCsabo]

Thanks! sansnom05 already explained it, and tetsuo asked me to create a ticket (https://sourceforge.net/apps/trac/mpc-hc/ticket/1090).

[djsolidsnake86]

rev 2881 can't move the video window by clicking on it
solutions?

[Ger]

And then there are the retarded-feeling default keys for seeking vs. frame-stepping (swap them?) or the audio delay using one of the largest sized key (add Ctrl, or better swap with Pns Inc/Dec Size). Most of these were requested countless times, but will never be applied because of people screaming at every little change.

Is there really a significant opposition out there to changing these? Like you say, the current default feels retarded. It seems like common sense that left/right arrow keys should be assigned to seeking, not just because that is the way almost every other player works, but also because I imagine almost every user out there uses seek more often than frame step.

I always use a reg file to assign medium jump (since key frame jump is splitter dependent) to the left/right arrow keys and add the CTRL modifier to frame step (so yes, effectively swapping them) when I install MPC-HC on a new computer. Personally, I don't feel strongly either way about the +/- keys, so I don't mind if they are changed like you describe either.

I encourage anyone with SVN access to go ahead and make the change, at least for the left/right keys, and then we'll see if there is much screaming about it afterwards. I doubt there are many people out there refusing to adapt to using CTRL for frame step.

[tetsuo55]

Discussing which keys should be used by default is a never ending story of various user opinions.
I'm open to changing them but such a thing cannot be done without good preperation, as suddenly keys that worked for 6 years will fail for upgrading users (possibly millions of them)

So what to do:
Create a list of proposed default key changes, for each change explain in detail why the new key is the best possible solution ever.

PS. I will not accept "everyone else uses that key" as the main argument, but i will take it as a 2nd one.

PPS: please also take this chance to ask for things that currently cannot be assigned/changed

[Mercury_22]

Discussing which keys should be used by default is a never ending story of various user opinions.
I'm open to changing them but such a thing cannot be done without good preperation, as suddenly keys that worked for 6 years will fail for upgrading users (possibly millions of them)

So what to do:
Create a list of proposed default key changes, for each change explain in detail why the new key is the best possible solution ever.

PS. I will not accept "everyone else uses that key" as the main argument, but i will take it as a 2nd one.

PPS: please also take this chance to ask for things that currently cannot be assigned/changed

Please change Play/Pause's default key cause it's messing up many other usual commands for left click like enter / exit full screen or drag windows or... also it's a big annoyance and even a show stopper for any NEW user since AFAIK the current usage for Left Click it's not used in any other player and it will fix play /pause problem in compact mode without the need for any new code = rev 2866 which has caused a new problem as you can reed in all this posts

My suggestion is to use Middle Down
It's also very easy you just need to change the "L" in AppSettings.cpp's line 208
208 ADDCMD((ID_PLAY_PLAYPAUSE, VK_SPACE, FVIRTKEY|FNOINVERT, IDS_AG_PLAYPAUSE, APPCOMMAND_MEDIA_PLAY_PAUSE, wmcmd::LDOWN));
To "M"
208 ADDCMD((ID_PLAY_PLAYPAUSE, VK_SPACE, FVIRTKEY|FNOINVERT, IDS_AG_PLAYPAUSE, APPCOMMAND_MEDIA_PLAY_PAUSE, wmcmd::MDOWN));:)

[Polcius]

@Polcius

If you post some measurement data here (http://forum.doom9.org/showthread.php?t=157722), I will try to draw some charts for you (2D CIE gamut view and TRC curves).
I need a spotread log file with XYZ values from uncorrected and corrected states measured in the MPC-HC window. You can use the AVS-HD test videos (http://www.avsforum.com/avs-vb/showthread.php?t=948496) to measure 10 gray-scale patches and the 3 primaries. (You should also clear your VGA LUT for the "uncorrected" measurements.)
Example script:
C:\Argyll\bin\spotread.exe -d -yl -db -H uncorrected.log

I will check myself with the ColorHFCR soft., but thanks for your help.

[Dstruct]

In DispcalGUI, I had to install the profile "system-wide" to see any change in MPC-HC with CMS enabled.

If I installed the profile "for only this user", CMS enabled in MPC-HC didn't seem to change anything.

Interesting. I already wondered why we can't choose a profile from within MPC directly.

Expecially for users with more than one screen this could be useful I think.

XnView has such feature for example: http://www.xnview.com

[jeremy33]

Hi,

Just a little question, is it normal I still have black screen with mpchc 2881 with catalyst 10.12 ?

- I have black screen with Full Float processing point.
- With Half Float processing point I have black screen with some part of the image in pink.
- Without Half Float processing and Full Float processing point the image is ok.

[Virtual_ManPL]

Are someone working on fixing system hanging and video stutter bugs when playing HD movies on Win 7 + EVR CP ?

It's big pain in the ass for user like me, who migrate from XP to 7 and see this... :p

[JanWillem32]

I'm testing mixer settings to try to solve the black screens problem on opening. I've disabled the "16-235" setting, because it harms quality and it uses the wrong levels for chroma. Also not working: D3D Fullscreen GUI Support and Force 10-bit RGB Input.
Currently the mixer can only do 8-bit, full range RGB. I'm trying hard to solve that. In the mean time, DXVA and many external codecs will correctly output 4:2:0 chroma, and some software codecs, like the internal ones are force-mixed to 4:2:2 (with the ugliest form of chroma blurring ever). Those who care to up-sample chroma, might sometimes have to switch between up-sampling from 4:2:0 in one occasion and from 4:2:2 in the other. Only the three EVR types are affected, the other renderers don't up-sample chroma by default.
The other thing is my attempt at temporal dithering. Later on dithering will probably need to become user-selectable, including the allowed noise levels. These builds use use a medium level of 8/256 or 8/1024.

[Casshern]

I'm testing mixer settings to try to solve the black screens problem on opening. I've disabled the "16-235" setting, because it harms quality and it uses the wrong levels for chroma. Also not working: D3D Fullscreen GUI Support and Force 10-bit RGB Input.
Currently the mixer can only do 8-bit, full range RGB. I'm trying hard to solve that. In the mean time, DXVA and many external codecs will correctly output 4:2:0 chroma, and some software codecs, like the internal ones are force-mixed to 4:2:2 (with the ugliest form of chroma blurring ever). Those who care to up-sample chroma, might sometimes have to switch between up-sampling from 4:2:0 in one occasion and from 4:2:2 in the other. Only the three EVR types are affected, the other renderers don't up-sample chroma by default.
The other thing is my attempt at temporal dithering. Later on dithering will probably need to become user-selectable, including the allowed noise levels. These builds use use a medium level of 8/256 or 8/1024.

MSVS 2010 versions only, based on v1.4.1.0
x86: http://www.mediafire.com/?vo59779487ea36n
x64: http://www.mediafire.com/?tial8e9aps1bwtz

Thanks for the new test version, unfortunately things are even worse. Now both YUY2 and NV12 produce only black screens. I never had problems with chroma upsampling with the Radeon 5970. With my previous card (2600 PRO) chroma upsampling was horrendous and i had to use the shader for decent results. As the Radeon chroma upsampling works well with the NV12 colorspace on the 5970 (tested catalyst 10.4-10.9) and as the pre 283x versions worked without black screens (which many people report) i am all for a revert until the new code is stable.

[cyberbeing]

There appears to be a bug which prevents enabling Color Management in the recent builds on WinXP SP3 x86. Once a video is loaded, 'Enable' is grayed out.

Last tested working build was 2833. It was broken in 2840, so one of the commits between 2834 and 2840 broke it.



@JanWillem32 and others working on MPC-HC CMS

I'm now realizing how badly we need an option for MPC-HC lcms to output with the same gamma as the profile.

If I calibrate to a 2.2, 2.35, or 2.4 power curve, all is well.

The problem is if I calibrate to a specialized gamma curve like a scaled Rec.709 gamma and want to keep that custom gamma after lcms has its way with it. As it currently stands, MPC-HC color management forcibly adapts my custom Rec.709 gamma to 2.2, 2.35, or 2.4 which completely defeats the purpose of calibrating to a specialized gamma curve. And just to be clear, adding a Rec.709 gamma option would result in the same problem because I use a Rec.709 gamma which is scaled to my particular viewing conditions. A native gamma option would be the only true fix.

Edit: I also seem to get strange gamma results when using XYZ LUT ICC profiles. Matrix profiles are fine.

[Dstruct]

There appears to be a bug which prevents enabling Color Management in the recent builds on WinXP SP3 x86. Once a video is loaded, 'Enable' is grayed out.

Last tested working build was 2833. It was broken in 2840, so one of the commits between 2834 and 2840 broke it.

Confirmed!

[JanWillem32]

@Casshern All chroma blurring in MPC-HC is CPU based, and only available in EVR. (VMR-9 users can also use shaders to correct chroma.) The up-sampling failures are from the external mixer that doesn't always select the correct type. There's only the exception with the internal codecs that are forced to the YUY2 type, instead of a proper 4:2:0 or 4:4:4 format as it in many cases should be. (The conversion to YUY2 is still done by the CPU.)
Black screens on starting of playback could be the mixer, as well. During testing it turned out that only very few formats actually worked. Later on I found out that the mixing is done though 8-bit, full range RGB in any case.
About reverting: I already reverted the input list for the mixer type selection, but there have been a lot of changes on the codecs and splitters side as well (not my specialty).

@cyberbeing I don't know anymore what and how to completely linearize the BT.601/BT.709 gamma. With Digital Cinema I know that it's gamma 2.6 out, 1/2.6 in, without exceptions. For BT.601/BT.709, I don't know what to use. There must be some reference how to do conversions to lossless video with BT.601/BT.709 encoding. Until I find out how it's done, I'm not changing the gamma shaders and any other gamma handling.
For using the native display gamma, the best thing I could think of is to linearize the gamma and then instruct lcms to use that 1.0 gamma on input (and trigger at least 16-bit surfaces to provide enough accuracy to bend the gamma curve in the first place).
The other things will require a review. Do Half and Full FPP get grayed out as well? The requirements of these three all depend on the support of floating point surfaces.

Maybe with some minor types added to the mixer playback works? If you see a non-RGB-type from the mixer with these builds, then there's something broken.

[Dstruct]

Do Half and Full FPP get grayed out as well? The requirements of these three all depend on the support of floating point surfaces.

Yes, Floating Point options are also grayed out on playback (rev 2860)!

[JanWillem32]

@Dstruct Here are two variants, one with lower requirements on the checking, one without the basic checking for floating-point surfaces support.

[Dstruct]

@Dstruct Here are two variants, one with lower requirements on the checking, one without the basic checking for floating-point surfaces support.
MSVS 2010, x86 only since this is for XP: http://www.mediafire.com/?by1b5pl84tb3hc7

Ok,

both versions don't have the floating point and color management options grayed out anymore.



But:

1. Color management doesn't seem to do anything (hardware calibrated NEC MultiSync LCD2490WUXi with ICC profile enabled in Windows XP) .Don't see any difference with it enabled or disabled.

2. Stats and Tearing Test don't work at all anymore (VRM-9 renderless, Nvidia GeForce 7600GS)

[JanWillem32]

Can you activate both floating point formats? (Requires restart of video to switch.)
This build has a half functional final shader, not recommended for regular use, but it's valuable to test.
You can test if the color management is working by switching between the 2.2 and 2.4 gamma functions, the gamma of the playing video should change quite a bit.
Stats and tearing test not working is odd. Are you sure that you are using 3D surfaces in the options\output tab?

[Dstruct]

Can you activate both floating point formats? (Requires restart of video to switch.)

Yes I can. But: The Stats always says HalfFP even if Full Floating Point processing is enabled. Limitation of my graphics card (Nvidia GeForce 7600GS)? With old 2833 revision I get FullFP displayed in Stats!


This build has a half functional final shader, not recommended for regular use, but it's valuable to test: http://www.mediafire.com/?363isaf24cqyza8
You can test if the color management is working by switching between the 2.2 and 2.4 gamma functions, the gamma of the playing video should change quite a bit.

Confirmed. It's working.


Stats and tearing test not working is odd. Are you sure that you are using 3D surfaces in the options\output tab?

Of course. Your build from http://forum.doom9.org/showpost.php?p=1473834&postcount=15949 didn't have this problem! The new mpc-hc half final.7z build also is showing Stats and Tearing Test properly again!

[Dstruct]

The new mpc-hc half final.7z build also is showing Stats and Tearing Test properly again!

Mmh, weird. Now it' not working anymore with this build (VMR-9 renderless, 3D surfaces, Bicubic 0.75)! Checkmarks are displayed in front of "Tearing Test" and "Stats" menu entries but I don't see anything on the screen!?

[cyberbeing]

@Dstruct Here are two variants, one with lower requirements on the checking, one without the basic checking for floating-point surfaces support.
MSVS 2010, x86 only since this is for XP: http://www.mediafire.com/?by1b5pl84tb3hc7

Unlike Dstruct color management is working in both of those builds.

With VMR9, Full-floating point works.

With EVR, only Half-floating point works. Use of Full-floating point produces distorted output, hangs the video driver, and eventually BSOD the entire computer.

Same behavior as above with the half functional final shader build.

[Dstruct]

Mmh, weird. Now it' not working anymore with this build (VMR-9 renderless, 3D surfaces, Bicubic 0.75)! Checkmarks are displayed in front of "Tearing Test" and "Stats" menu entries but I don't see anything on the screen!?

Got it:

Only happens with Full Floating Point processing enabled. With Half Floating Point processing I get Stats and Tearing display.

So FullFP obviously isn't working anymore.

[cyberbeing]

Not seeing that here, Dstruct.

FullFP works with VMR9, as well as Color Mangement, Stats with all three test builds.

FullFP is broken in all three test builds with EVR-CP. EVR-CP & HalfFP works with Color Management and Stats in all three test builds.

In MPC-HC 2833, everything works including FullFP in EVR-CP.

This is with a 7800 GTX 512.

[JanWillem32]

@Dstruct The Nvidia GeForce 7600GS can't do full floating point processing. I'll add a check for it when I have the time. Can you test a simple shader, like invert on a 2.0, 2.0a and a 3.0 profile (not the software emulations)? I think your videocard might not like shaders in 2d mode.

[Dstruct]

@Dstruct The Nvidia GeForce 7600GS can't do full floating point processing.

Alright. Didn't know that.


I'll add a check for it when I have the time.

This would be great, thanks!

Can you test a simple shader, like invert on a 2.0, 2.0a and a 3.0 profile (not the software emulations)? I think your videocard might not like shaders in 2d mode.

What do you mean with 2.0, 2.0a and 3.0 profile? "Invert" pixel shader is working fine in "mpc-hc half final.7z" build, but only with "3D surfaces" enabled. "2D surfaces" and "Regular offscreen plain surface" both give non-working shaders.

[JanWillem32]

View -> Shader editor -> select Invert from the drop-down list -> select different profiles in the top-right drop-down list while playing a video.

[Dstruct]

View -> Shader editor -> select Invert from the drop-down list -> select different profiles in the top-right drop-down list while playing a video.

Ok, thanks for explaining.


Result:

Only the "sw" versions don't work at all ("Could not load shader: invert" displayed in statusbar). All others work fine!

[JanWillem32]

Added extra checks, added 10-bit out support for VMR-9.
It could still give some trouble with the final pass shader (color management and dithering).

[Dstruct]

Added extra checks, added 10-bit out support for VMR-9: http://www.mediafire.com/?9q32jll39z33v4i

I guess this also isn't supported by the Nvidia 7600GS (because the option gets grayed out on playback)?

[cyberbeing]

That build broke VMR-9 FullFP for me (grayed out).

[JanWillem32]

@Dstruct: I checked, indeed no support for the 10-bit format.
@cyberbeing: Can you try to reset your settings and then activate it again? It should only gray out after starting rendering. I can't find any flaws in the code (the lines for FP16 and FP32 are in nearly all cases next to each other, and nearly identical).

[cyberbeing]

FullFP grays out after starting rendering, but even if enabled beforehand, it disables itself (as seen in stats).
http://img541.imageshack.us/img541/8251/fullfp.png

HalfFP works and never grays out, even after rendering starts.
http://img837.imageshack.us/img837/8757/halffp.png

The 'half final' and 'checkadapt' builds didn't have this issue. This only occurs with the 'hfsplit' build.

[Dstruct]

@Dstruct: I checked, indeed no support for the 10-bit format.

Alright, thanks! Btw: Where do you get that information from?