Media Player Classic - BE Win32/x64

[FuGGeroo]

I just did a clean install of Windows 10.

Reinstalled MPC-BE, went to Options and discovered "Store settings in player folder" is greyed out and un-clickable.

http://s1.postimg.org/t0l2c7v67/sett.png

[Aleksoid1978]

I just did a clean install of Windows 10.

Reinstalled MPC-BE, went to Options and discovered "Store settings in player folder" is greyed out and un-clickable.

http://s1.postimg.org/t0l2c7v67/sett.png

I think you install in "Program Files" folder. Store in this folder requeried administrative rights.

[FuGGeroo]

I think you install in "Program Files" folder. Store in this folder requeried administrative rights.

But I am an admin on my PC.
I installed it in C:\Program Files (x86).

So, this is how its supposed to work in Windows 10?
Or did I accidently change something in Windows 10 that made it act this way?

Edit: I fixed it. W10 is such a pain...

[v0lt]

@NikosD
What sample?

[NikosD]

@NikosD
What sample?
ANY VC-1/WMV3 sample, crashes the DXVA1 VC-1 decoder.

[edigee]

By using the good old method of "trial and error" I managed to isolate the error.

For my system:
WinXP SP3+
AthlonXP 2600+
Radeon 3650 - Latest Catalyst for AGP 14.4

the crash appears from MPC-BE.1.3.1.1.4477.x86 version and later

The last working version for DXVA VC-1 is MPC-BE.1.3.1.1.4469.x86

Excuse me ,but how you manage to have any DXVA enable on your system considering that the old Athlon doesn't have SSE 2 instructions? AFAIK SSE2 is mandatory for DXVA to work.

[NikosD]

AFAIK SSE2 is mandatory for DXVA to work.

That's an old myth floating around, probably because of OEM's drivers or OEM's commercial SW or Microsoft, I don't know, that is simply not true.

The artificial requirement of SSE2 presence globally for various applications including OSes (like Win 8.x or later), is more and more common, though.

[beter]

Fixed.

Thanks.
And this ts file:
https://www.sendspace.com/file/u9ah8b

The player crashes when it tries to open this file.

[edigee]

That's an old myth floating around, probably because of OEM's drivers or OEM's commercial SW or Microsoft, I don't know, that is simply not true.

The artificial requirement of SSE2 presence globally for various applications including OSes (like Win 8.x or later), is more and more common, though.

Well it's good to know ,because I had almost the same system (Athlon 2400 XP ,HD 3450 and finaly a 3650 AGP ,XP SP3) and I have never ,never been albe to have any kind of DXVA. It's true ,that's hapened a few years ago.

[Aleksoid1978]

Thanks.
And this ts file:
https://www.sendspace.com/file/u9ah8b

The player crashes when it tries to open this file.

Thanks - fixed. :)

[FuGGeroo]

Is it possible to right click on a video in File Explorer, and select add to MPC-BE's playlist?

[Aleksoid1978]

Is it possible to right click on a video in File Explorer, and select add to MPC-BE's playlist?

http://i.imgur.com/dQEVQqF.png

[73ChargerFan]

MPC-BE 1.4.5 (build 699) beta - x86 (https://yadi.sk/d/207pBuuTicF22), x64 (https://yadi.sk/d/1OaBDAqAicF26).
x64 link is down - the service won't let me download

[NikosD]

@NikosD
What sample?
Hi.

Any news on that ?

Was it possible to reproduce ?

Do you think is an easy fix ?

Thanks.

[v0lt]

x64 link is down - the service won't let me download
Try http://mpc-be.org/builds/

Was it possible to reproduce ?
No. We do not have available a processor without SSE2.

[FuGGeroo]

http://i.imgur.com/dQEVQqF.png

I've done that but when I right-click on a video in File Explorer, its not there...

Edit: It only shows up for folders/directories containing video files in File Explorer. When I right-click on an individual video, its not there.

[NikosD]

No. We do not have available a processor without SSE2.

So, my initial thought of being a SSE2 fault is probably right.

But then again, why DXVA1 H.264 works like a charm with my SSE CPU even using latest nightly build ?

Is it possible for you to recompile DXVA1 VC-1 with SSE support even if you don't have a SSE CPU, in order for me to test it ?

[Aleksoid1978]

So, my initial thought of being a SSE2 fault is probably right.

But then again, why DXVA1 H.264 works like a charm with my SSE CPU even using latest nightly build ?


About DXVA 1 - VC-1/MPEG2 code "store" at ffmpeg's side(use my custom code) and use gcc to compile, but H.264 code store in project's .cpp/.h files and compile with VS.

What about software H.264/VC-1/MPEG2 decoding ??

P.S. Try this build (https://yadi.sk/d/igiSj5SHihxb7)

[huhn]

wasn't dxva2 not possible on windows xp?

http://forum.doom9.org/showpost.php?p=1563106&postcount=9647

you may trick around with dotnet 3.5 and EVR at least i got DXVA working in windows xp x64 with EVR which shouldn't work too. but yeah it's a lot of years ago and to be honest there is a time old hardware is just that old hardware.

the old HD 3xxx hardware acceleration is terrible anyway.

[NikosD]

About DXVA 1 - VC-1/MPEG2 code "store" at ffmpeg's side(use my custom code) and use gcc to compile, but H.264 code store in project's .cpp/.h files and compile with VS.

What about software H.264/VC-1/MPEG2 decoding ??

Hmm...I see...So it looks like GCC does the bad job by enabling by default the SSE2.

But I tried all SW decoders (VC-1/WMV3, H.264, MPEG2) and work fine without crashes, even with latest nightly.

Of course they are slow to death for AthlonXP.

Did you take a look at that build specifically, that started crashing badly DXVA1 VC-1 ?

ALL VC-1/WMV3 samples crash with DXVA1 VC-1 after that specific build, but not with all the previous.

Maybe there is something specific to that build, because all SW decoders work fine and DXVA1 H.264 too.


wasn't dxva2 not possible on windows xp?

http://forum.doom9.org/showpost.php?p=1563106&postcount=9647

the old HD 3xxx hardware acceleration is terrible anyway.

You are really obsessed commenting my posts, although your comments are (almost) always irrelevant or wrong.

As I wrote a few times above, we are talking about DXVA1 supported by WinXP and not DXVA2 unsupported by WinXP.

And HW acceleration of Radeon 3000 series is perfect for BluRay Spec H.264/VC-1 clips, btw.

[huhn]

You are really obsessed commenting my posts, although your comments are (almost) always irrelevant or wrong.

As I wrote a few times above, we are talking about DXVA1 supported by WinXP and not DXVA2 unsupported by WinXP.

And HW acceleration of Radeon 3000 series is perfect for BluRay Spec H.264/VC-1 clips, btw.

i'm not the first person you said that about. not a nice thing to tell or think about people.

DXVA isn't suppose to work with EVR in XP but with dotnet 3.5 i got it working back in the time. could be xp 64 that made it possible. could be DXVA2 or couldn't be DXVA2. give it a try or ignore it this is your choice.

and according to MSDN EVR it is DXVA2:
https://msdn.microsoft.com/en-us/library/windows/desktop/cc307941%28v=vs.85%29.aspx

[NikosD]

WinXP is using VMR for DXVA1 HW acceleration.

[huhn]

yeah i know and that's my point. i was able to use EVR which shouldn't work only with VMR and other older renderer.

but there was an issue with AMD/ATI and DXVA2 back in the days that mpeg2 decoding didn't work with DXVA2 but with DXVA1 (was the case with my 6770). but i can't say for sure if DXVA2 was used anyway.

as far as i remember i just needed dotnet framework 3.5 and EVR with DXVA worked with my nvidia ion on XP 64.

i can't deny that there is a high possibility that this isn't working at all.

[nevcairiel]

Its really up to the video driver if DXVA2 works. There are a few drivers which allow limited formats in DXVA2 on XP, but its usually half-broken or limited to one of the common formats, or something. Not really worth calling "supported" in any sense of the word. More like luck.

[huhn]

i see.

[ryokoseigo]

x64 version wont load for me... I'm running windows 10 if that matters, but I haven't been able to run the x64 version. It doesn't even appear to go into the task manager... I've tried compatibility mode etc, and still nothing.

Does anyone know of a means to fix this? I have tried the latest release and even an older release 1.3 to no avail.

[v0lt]

I have tried the latest release...
What is the full build number?

[Dlget]

x64 version wont load for me... I'm running windows 10 if that matters, but I haven't been able to run the x64 version. It doesn't even appear to go into the task manager... I've tried compatibility mode etc, and still nothing.

Does anyone know of a means to fix this? I have tried the latest release and even an older release 1.3 to no avail.

i'm useing win10 x64 with no issue & mpc-be works fine(using nightly & stable also works fine)
Try clean installing win10.(format then install from scratch)

[Dark Eiri]

i'm useing win10 x64 with no issue & mpc-be works fine(using nightly & stable also works fine)
Try clean installing win10.(format then install from scratch)

Same. Never had any issue. Not with the old Technical Previews, not with the public 10240 version, not with the new insider builds.

[FuGGeroo]

i'm useing win10 x64 with no issue & mpc-be works fine(using nightly & stable also works fine)
Try clean installing win10.(format then install from scratch)

I have two PCs with clean installed W10 Pro x64 running Core i7 47XX CPUs with HD4600 and I'm getting random crashes, stuff not working right...

[Klaus1189]

updated german Translation ;)


I have some trouble with downloaded YouTube-videos:

The problem only occurs if the framerate of the video is 23.976 fps:

MPC-BE:
dash: 23.980 -> maybe also take a look at this?
muxed mkv: 23.974 -> ??

MPC-HC:
dash: 23.976 -> fine
muxed mkv: 23.974 -> ??


I download the dash files which should be fine with 23.976, but if I mux the video and audio file together with MKVToolNix, then MPC-BE displays them as 23.974 in display stats (CTRL + J) and I get more stutters. See attachment.

Muxing doesn't change the videostream, so I'm helpless what's going on. I want MPC-BE recognizing it as 23.976 (24.000/1.001) like the files "hopefully" are. I'm pretty sure.

But keep in mind there are some "old" files still on youtube which are "true" 23.974 regardless what you do. So take a look at a newer video.

[v0lt]

updated german Translation ;)
Thank you, but it's better to use a file storage services, instead of the forum.

MPC-BE:
dash: 23.980 -> maybe also take a look at this?
muxed mkv: 23.974 -> ??
If you give me both files, you'll see.

[NikosD]

P.S. Try this build (https://yadi.sk/d/igiSj5SHihxb7)

Sorry for being late to my respond, but I was ready to make another comment about the issue, when I saw that you had updated your post with a link to a new build that I wasn't aware of.

Anyway, I tried that build just now and unfortunately it crashes exactly like all the others from 1.3.1.1.4477.x86 and above.

[Klaus1189]

@ v0lt:
OK, can I use Google Drive for future translation updates?
Sent you links to the files.

[v0lt]

OK, can I use Google Drive for future translation updates?
Yes, it would be better.

Sent you links to the files.
Typical file:
DefaultDuration = 41708333
AvgTimePerFrame = 41708333 / 100 = 417083
FPS = 1000000 / 417083 = 23.976

Your file:
DefaultDuration = 41711111
AvgTimePerFrame = 41711111 / 100 = 417111
FPS = 1000000 / 417111 = 23.974

[Klaus1189]

That's exactly the point. Which File of the two is reported right? The File has only been muxed and then it is reported differently. Before 23.980 and then 23.974. MPC-HC reports 23.976 on the dash file as already mentioned.
Does that difference cause another playback of the file? Why is the dash file played fine on MPC-HC?
I'm clueless ...

[v0lt]

@Klaus1189
I think the problem came from the strange parameters of source MP4 file.
MediaInfo
Frame rate mode : Variable
Frame rate : 23.976 fps
Minimum frame rate : 23.974 fps
Maximum frame rate : 1 046.512 fps
In any case, the MKV parameter DefaultDuration merely informational. And it should not affect the playback quality.

[Klaus1189]

Tested a long 23.974 file with my HTPC (Intel HD 4600). Seems to don't bother for playback. Do newer nvidia cards support 23.976 like Intel hd46000 or still 23.972?

[huhn]

only intel had this issue till haswell known as intel 24p bug and it was an chipset issue.

23.972 is known as the small 24p bug.

my HD 4400 has a clock of 23.97600-23.97605 thsi is a pretty perfect clock but that doesn't really help because there are more than 1 clokc that needs to be perfect in the first place.

never heard of an problem with nvidia and nvidia had custom resolution support for a very very long time too.

[Klaus1189]

I know that but there is a bug resulting only available for 24hz and not for 23 hz setting.
The Intel HD 24hz bug. Omg very old. Playing 23.976 as 24.000 very bad. INTEL HD 4600 is the Best you can use for that.

[Klaus1189]

I already contacted Mediainfo developers some years ago. They told me that this has to do with the first and the last frames of the video. For mp4 container it only uses the middle frames, but for this dash container with mp4 extension it includes all frames resulting in variable framerate vbr which is not true. They told me to spend money to get this fixed.
Can you check these dash files with special tools. Maybe you know something, I don't.
And if you mux the videostream of the dash file to a mkv, MediaInfo tells us 23.976 cfr.
Quite strange thing, but I think we should check this very deep.

[Klaus1189]

"The custom timing feature is usable, but not without its quirks. Adding a custom resolution is straightforward. Setting the vertical parameters to values similar to the ones in the screenshot above achieves desired results, but the 23 Hz resolution gets saved as 24 Hz."
Source: http://www.anandtech.com/show/5969/zotac-geforce-gt-640-review-/5

I tried that also, but same happened.

If the Information is not up to date let me know. I want to learn ;)
Is there any better than hd4600? Still 23.972 or have the New nvidias true 23.976?
Would be nice to know that. Otherwise i must buy one test and send back which i don't want to do.

[huhn]

of cause they have proper 23p.

but even with the clock from my HD 4400 i get a repeated frame every ~13 min.

you can buy 10 gtx 960 all and they will not have the same clock same for onboard sound/sound cards.

aiming for the theoretical perfect clock of 23.97602 hz doesn't really help at all. you have to match the reference clock with is usually the audio clock and it is very unlikely to be 23.97602hz.

[salam2009]

Why does mpc play only audio of h.265 files? I'm on latest version of k-lite mega codec pack!

[Dlget]

MPC-BE crashing on a video.
Here is the dump

[Aleksoid1978]

MPC-BE crashing on a video.
Here is the dump

What version MPC-BE ?? upload a video, or short sample. But - don't use forums - use external file uploaders.

[NikosD]

Sorry for being late to my respond, but I was ready to make another comment about the issue, when I saw that you had updated your post with a link to a new build that I wasn't aware of.

Anyway, I tried that build just now and unfortunately it crashes exactly like all the others from 1.3.1.1.4477.x86 and above.
Anything new ?

Is there someone looking on this ?

TIA

[Klaus1189]

Why does mpc play only audio of h.265 files? I'm on latest version of k-lite mega codec pack!

Are you sure you're in the right forum?
MPC-BE in K-Lite?
Anyway. I recommend not using K-Lite.

Better you use MPC-HC with integrated LAV Filters
https://mpc-hc.org/
https://nightly.mpc-hc.org/

Or you can use MPC-BE with internal Filters and if you want set ffdshow or LAV Filters as external Filters. But don't have to, the internal are very nice.

Then we can talk here ...

[Klaus1189]

Yes, it would be better.


Typical file:
DefaultDuration = 41708333
AvgTimePerFrame = 41708333 / 100 = 417083
FPS = 1000000 / 417083 = 23.976

Your file:
DefaultDuration = 41711111
AvgTimePerFrame = 41711111 / 100 = 417111
FPS = 1000000 / 417111 = 23.974



I checked the muxed MKV with MediaInfo in advanced mode under debug:
.
.
.
Duration : 225017
Duration : 3mn 45s
Duration : 3mn 45s 17ms
Duration : 00:03:45.017
Duration : 00:03:45.017 (00:03:45;01)
Frame rate : 23.976
Frame rate : 23.976 fps
Frame count : 5395
.
.
.

If I calculate the time 225017ms diveded with number of frames 5395, then I get 41,708433734939759036144578313253ms
then 1/x and then x 1000
--> 23,97596626032699751574325495407
which is quite exact 23.976 and not 23.974

[v0lt]

@Klaus1189
Your method will not work for VFR video. In addition, it requires the analysis of the entire file.

DefaultDuration parameter recorded in the header of your file.