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

[Underground78]

Hmm, I should have tell you, there is a bug in the website that prevent you from seeing some builds ... :/ Put "Количество строк:" on 25 if you want to see the rev between 3268 and 3337.

Also does this happen with the x86 build too ?

[JohnLai]

Anyway..... I only tested 3 x86 builds from xvidvideo's builds....on 64bit win7 SP1.
3268 = perfectly fine.
3337 = BEX
3509 = BEX -->no '64' wording

Well...another results from xhmikosr's builds 64bit:
3280 = no problem
3285 = no problem
3291 = no problem
3297 = no problem
3322 = no problem
3324 = BEX64
3326 = BEX64

So...somewhere beween 3322 and 3324...something messed up, no idea what but changelog is :
Revision 3324 - Directory Listing
Modified Tue Jul 5 11:38:59 2011 UTC (3 weeks, 4 days ago) by aleksoid

Fix : x64 build crash if in Options try to use "WinLIRC" or "uICE" and press "Cancel".

Revision 3323 - Directory Listing
Modified Tue Jul 5 06:05:53 2011 UTC (3 weeks, 4 days ago) by aleksoid

Ticket #1493, thanks to judelaw.

Revision 3322 - Directory Listing
Modified Tue Jul 5 00:16:05 2011 UTC (3 weeks, 4 days ago) by aleksoid

Fix : MPEGSplitter, breaks standalone filter.

[Underground78]

Can you try this test build (http://www.mediafire.com/?7rx0pabsqq0e780) which has r3323 reverted and tell me if it helps ? Also try this debug build (http://www.mediafire.com/?i8myk3ril0l31en) and tell me if you get some useful informations when it crashes.

[JohnLai]

The results:
Test build = BEX
Debug build = no problem, I think? (Since no errors come out when exiting D3D)

[Underground78]

It seems the problem was not introduced where you think it was ... Are you really sure the problem never happens with r3322 from xhmikosr's website ?

[JohnLai]

Yup, I double checked again with 3322 and 3324 (64bit and 32bit) from xhmiskosr.

[Ayukawa]

Concenring the problem of DXVA not working on a laptop with ATI/Intel switchable graphics mentioned on the previous page, the last version that works on xvidvideo.ru is 1.5.1.2925.x64. After that the video will not play on the ATI chipset. Thanks.

[Underground78]

Yup, I double checked again with 3322 and 3324 (64bit and 32bit) from xhmiskosr.

It seems extremely doubtful that rev 3324 introduced the problem so if it's not 3323, I'm running out of ideas.

Here what Tetsuo told me on IRC tonight:
<tetsuo--> Underground78: that crash johnylai has
<tetsuo--> Underground78: its caused by his: driver, virusscanner, filter or trojan/virus
<tetsuo--> Underground78: tell him to update his videocard drivers and virusscanner, if that doesnt help tell him to only use internal filters
<tetsuo--> i cannot log in here, so i cant tell him

Concenring the problem of DXVA not working on a laptop with ATI/Intel switchable graphics mentioned on the previous page, the last version that works on xvidvideo.ru is 1.5.1.2925.x64. After that the video will not play on the ATI chipset. Thanks.

What is the first non working version you have found on xvidvideo.ru ?

[nimfe]

Anyone noticed that using MPC-HC and Logitech Gaming Software (older or 8.0.127) with 'Bicupic A=-0.60 PS2.0)' or 'Bicupic A=-0.75 PS2.0)' resizer option just result garbage output? Curiously other options dont..

In non current Logitech Software you could just disable MPC-HC but in newer one it wont help. Now I know logitech software is garbage as general but there is no real way use gaming keyboards like g150/g19 without it.

Perhaps option to disable keyboard lcd interface would be in order? Ofcoz it does not make sense that only those 2 shaders fail but i leave that to coders. :)

Non zoomed: http://img8.imageshack.us/img8/9933/36nonzoom.png
Fullscreen (any other zoom% does same though): http://img836.imageshack.us/img836/5599/34fullscreengrab.png

Mpc version: 1.5.2.3456 64bit (and yes other versions dont work either), EVR-sync.
Other stuff: W7-64, NVidia Gtx460 with 280.19, Asus M4A79T with Phenom2 x4 955, G510 keyboard and uptodate drivers.

[rahzel]

I have an ATI 5770 and use EVR CP, is the "YV12 Chroma Upsampling" shader still needed?
What is the difference between pre-resize and post-resize pixel shaders? I tried both but I hardly see any difference, post-resize seems slightly better (I use Sharpen Complex 2 and YV12 Chroma Upsampling). Which one should I use?

I've asked this several times and never have gotten an answer. From what I've read if you use say ffdshow video decoder and set it to output RGB32 and have High Quality YV12 to RGB Conversion and Dithering checked, then the chroma upsampling bug won't affect you.

For sharpening, you want to generally use the post-resize pixel shader. For YV12 chroma upsampling I don't know if you should use post or pre-resize pixel shaders.
Using leeperry's provided test pattern, I just compared MPC's YV12 chroma upscaler shader and ffdshow's high quality YV12 -> RGB conversion. For me, putting the YV12 shader in pre-resizer provided better results -- comparable to ffdshow. There are other alternatives to chroma upsampling (madVR, custom shaders), but considering you have an ATI card, if you enable full/half floating point processing or 10-bit RGB output, you need to enable that shader (or some sort of chroma upsampler), otherwise MPC will use nearest neighbor and it looks very bad (lots of aliasing and blocking). You really have to zoom in pretty good to see the difference, though.

[Ayukawa]

What is the first non working version you have found on xvidvideo.ru ?
It's 1.5.1.2928.x64. The official 1.5.2.3456 and the latest revision I tried last night also don't work. Thanks.

[JohnLai]

Driver = Ati Catalyst 11.6b and 11.8 preview, HD5450
Antivirus = Comodo Internet Security
I only use internal transform filter H264/AVC (DXVA) decoding for H.264 (I disabled all source and transform filters except H.264 DXVA) , ffdshow 3949 64bit, haali media splitter. Later I will test to use only internal filters.

[betaking]

Driver = Ati Catalyst 11.6b and 11.8 preview, HD5450
Antivirus = Comodo Internet Security
I only use internal transform filter H264/AVC (DXVA) decoding for H.264 (I disabled all source and transform filters except H.264 DXVA) , ffdshow 3949 64bit, haali media splitter. Later I will test to use only internal filters.

Ati Catalyst 11.6b and 11.8 preview is not good driver for ati 5xxx!

[ryrynz]

With all the BE builds on Xvideo.ru that I've tried jumping is broken. The standard builds available from Xvideo work perfectly. The jumping problems seem to only occur with the small and medium jumps. Can someone pass that along?

[JohnLai]

Ati Catalyst 11.6b and 11.8 preview is not good driver for ati 5xxx!

:confused: Thus, enlighten me on which version of Catalyst that is a good driver for 5000 series and I shall retest again once I have the time.


Btw, Underground78, why your debug build didnt crash when I tested it?

[iron2000]

With all the BE builds on Xvideo.ru that I've tried jumping is broken. The standard builds available from Xvideo work perfectly. The jumping problems seem to only occur with the small and medium jumps. Can someone pass that along?

I too experienced jumping problems with the BE builds.
Instead of jumping forward, it jumps back to the time when the jump is initiated.

Haven't been using the standard build for a long time now.

[ryrynz]

I too experienced jumping problems with the BE builds.
Instead of jumping forward, it jumps back to the time when the jump is initiated.

Haven't been using the standard build for a long time now.

Yeah that's exactly it, it doesn't happen all the time though.
I'd also rather use the BE builds (The original is just plain ugly) why that mod hasn't been incorporated into the standard builds I don't know, but first things first the jumping needs to be fixed.

[ryrynz]

:confused: Thus, enlighten me on which version of Catalyst that is a good driver for 5000 series and I shall retest again once I have the time.


Btw, Underground78, why your debug build didnt crash when I tested it?

Use 11.7

[JohnLai]

No luck, same result with 11.7.

After some testing with 3514:
If display the statistics (ctrl + j) while exiting = BEX64 StackHash_1dc2 = not always, sometimes only
Not sure what condition, sometimes gets APPCRASH
If OSD is displayed (such as fast forward etc) while exiting = BEX64 StackHash_26e5

Ignore above, still 3514:
Only 3 random errors occured
BEX64 StackHash_26e5 (always)
BEX64 StackHash_1dc2 (rare)
APPCRASH (often)

[Underground78]

Btw, Underground78, why your debug build didnt crash when I tested it?

No idea and that makes difficult to find the bug ...

It's 1.5.1.2928.x64. The official 1.5.2.3456 and the latest revision I tried last night also don't work. Thanks.

Ok so that's r2928 which breaks it, I'm quite sure of that but I'm not sure how to fix it. This rev disables DXVA for some Intel chips like Sandy Bridge as it made MPC-HC crash. I will ask Casimir about that.

[ceb]

Using leeperry's provided test pattern, I just compared MPC's YV12 chroma upscaler shader and ffdshow's high quality YV12 -> RGB conversion. For me, putting the YV12 shader in pre-resizer provided better results -- comparable to ffdshow. There are other alternatives to chroma upsampling (madVR, custom shaders), but considering you have an ATI card, if you enable full/half floating point processing or 10-bit RGB output, you need to enable that shader (or some sort of chroma upsampler), otherwise MPC will use nearest neighbor and it looks very bad (lots of aliasing and blocking). You really have to zoom in pretty good to see the difference, though.
I have not enabled full/half floating point processing or 10-bit RGB output, I don't even know what they do. :p
You mentioned "nearest neighbor", is it connected to the output resizer options? I have it at bicubic A=-1.00.
Anyway since you have to zoom in a lot to see the difference I should be fine without it.

[misha0209]

hi all,
if this was discussed, probably was but i do not have the time to look into 894 pages.
so, my question is regarding EVR-CP and EVR-SYNC.
can someone explain to me which one i should be using? what is the difference?
in EVR-CP i get no dropped frames. in EVR-SYNC i get dropped frames each time i change my position in video timeline. is EVR-SYNC experimental? is it recommended over EVR-CP, for a 60Hz screen?
thanks!

[Mercury_22]

Can someone please add a "Key" / keyboard shortcut for "After Playback"-> "Once"-> "Shutdown" cause it defeats the purpose of having this commands since we can't use them from a Remote control
I'm not "familiar" with MPC-HC's code but it doesn't seem that much work to me and the benefits are well worth the "Cost" :) :confused:

P.S. See Ticket 1155 (http://sourceforge.net/apps/trac/mpc-hc/ticket/1155)

[Hera]

Anyone get,
D3DFS + No Audio Codec = No ability to interact with MPC:HC

Without D3DFS, a popup informs you about some gibberish with audio codec. File had Vorbris audio.

[rahzel]

I have not enabled full/half floating point processing or 10-bit RGB output, I don't even know what they do. :p
You mentioned "nearest neighbor", is it connected to the output resizer options? I have it at bicubic A=-1.00.
Anyway since you have to zoom in a lot to see the difference I should be fine without it.
to me, the biggest reason to enable 10-bit output/ffpp/hfpp is so that MPC can enable dithering, which helps with banding. if you look at a smooth grayscale ramp, you may notice banding, in which one of the aforementioned settings can help.

banding is introduced because most video cards only output RGB 444 0-255 while video content is usually YCbCr420 16-235, so both the luma and chroma have to be upsampled. dithering really smooths out gradients. this is easier to notice than fixing chroma upsampling, but other than looking at a grayscale ramp, again, the difference isn't that noticeable when watching a video.

there are a few things that need to be scaled when playing videos on a PC. the scaler in the MPC output settings is for resizing videos to your display AFAIK.

[pankov]

Guys,
I have a problem with VSFilter and more precisely with it's ability to delay the subtitles and even more precisely the way ZoomPlayer controls it in real time to change the delay value. I guess ZoomPlayer uses some kind of interface to do it but I'm not sure.
The latest version that I've found on xvidvideo.ru to be working is 1.5.2.3251 and the next one there (1.5.2.3314) is broken. Sadly there are ~60 versions in between but I can't find any of them so I can't narrow it down more.
Also I apologize for not being able to explain the problem more technically but I hope you'll be able to understand.

I've added a ticket to tracker
http://sourceforge.net/apps/trac/mpc-hc/ticket/1585
and I hope it will be easy to fix it.

[ryrynz]

Pankov if you head here

http://www.xvidvideo.ru/media-player-classic-home-cinema-x86-x64/stranitsa-2.html

and also click on page 1 at the bottom of the page you can access more builds in the range you specified.

[tetsuo55]

No luck, same result with 11.7.

After some testing with 3514:
If display the statistics (ctrl + j) while exiting = BEX64 StackHash_1dc2 = not always, sometimes only
Not sure what condition, sometimes gets APPCRASH
If OSD is displayed (such as fast forward etc) while exiting = BEX64 StackHash_26e5

Ignore above, still 3514:
Only 3 random errors occured
BEX64 StackHash_26e5 (always)
BEX64 StackHash_1dc2 (rare)
APPCRASH (often)Hi, I discussed this further with some others and checked on MSDN.
The advice they gave was(in order of what to try):

1. Update all your drivers to latest
2. Run anti-spyware, anti-virus on the whole system
3. Format and reinstall windows (making sure your computer does not get infected over LAN/WAN, connect it to the network after installing virusscanner/firewall)
4. Replace the PSU
5. Replace the RAM

Good luck.

[tetsuo55]

Wouldn't it be awesome if windows and mpc-hc where THIS stable: http://www.fastcompany.com/magazine/06/writestuff.html?page=0%2C0

[Uxorious]

It seems MPC-HC is having problems with 7.1 AAC content.
(I've had the problem forever, but currently I'm on v1.5.3.3474 as shipped in K-Lite Mega Codec Pack 7.5.4).

If I play a 7.1.m4a file in MPC-HC, I get this constant scratchy sound (I can hear what it is, but it's distorted).
Take the same file and convert it to a WAV (with Nero), and MPC-HC plays it fine.

Anybody else with this problem, or is it likely to be something on my end?

[ryrynz]

That's an amazing read Tetsuo, highly recommend it to everyone.

[JohnLai]

Dear tetsuo55, I just formatted my pc around 1 week ago and the problem still remains same. I already make sure all of my drivers are up to date. Done checking for virus stuff by uninstalling comodo and scans using Avira today, nothing detected.

RAM tested using Memtest86, no problem.
I got two PSU : Silverstone ST50F and Gigabyte GE-N400-C2 = same result.

Those errors only occur when exiting from D3D mode. If D3D mode is not used, it exited fine.
So, I am entirely unsure why errors happens with ATI card after version 3322.

Just in case, I tested with my Nvidia 8400GS using 280.19 drivers and it works wonder, no error at all.

[Underground78]

If it doesn't happen with all computers, it will make the problem really difficult to understand ... :(

It seems MPC-HC is having problems with 7.1 AAC content.
(I've had the problem forever, but currently I'm on v1.5.3.3474 as shipped in K-Lite Mega Codec Pack 7.5.4).

If I play a 7.1.m4a file in MPC-HC, I get this constant scratchy sound (I can hear what it is, but it's distorted).
Take the same file and convert it to a WAV (with Nero), and MPC-HC plays it fine.

Anybody else with this problem, or is it likely to be something on my end?

You should probably open a ticket on Trac and upload a small 7.1 AAC sample.

[pankov]

Pankov if you head here

http://www.xvidvideo.ru/media-player-classic-home-cinema-x86-x64/stranitsa-2.html

and also click on page 1 at the bottom of the page you can access more builds in the range you specified.

Thanks, ryrynz.
I'm sure I did look for versions on different pages but I guess either I've somehow missed them or they simply were not present the other day. Nonetheless I managed to narrow it down between 1.5.2.3262 which is OK and 1.5.2.3268 which is broken.
Looking at the changelog I'm pretty sure it's commit 3264 that broke it
Add two new options in VSFilter GUI like it is done in MPC-HC:
- "Sub pictures to buffer" (default to 4)
- "Allow animation when buffering" (enable by default).

[Underground78]

I don't think this commit really broke something, it just triggered an existing bug. Can you try playing a bit with the options ? I would guess that if you set "Sub pictures to buffer" to 0, I will work again.

[pankov]

Underground78,
I've tried that but it didn't help.
I've also tried unchecking the "Allow animation when buffering" checkbox but again no success.
:(

[tetsuo55]

Dear tetsuo55, I just formatted my pc around 1 week ago and the problem still remains same. I already make sure all of my drivers are up to date. Done checking for virus stuff by uninstalling comodo and scans using Avira today, nothing detected.

RAM tested using Memtest86, no problem.
I got two PSU : Silverstone ST50F and Gigabyte GE-N400-C2 = same result.

Those errors only occur when exiting from D3D mode. If D3D mode is not used, it exited fine.
So, I am entirely unsure why errors happens with ATI card after version 3322.

Just in case, I tested with my Nvidia 8400GS using 280.19 drivers and it works wonder, no error at all.Many of use ATI, i cannot reproduce the probleon my ATI.

Can you try this build? http://www.mediafire.com/download.php?9ap6i7kizkwj0a7

[Underground78]

Underground78,
I've tried that but it didn't help.
I've also tried unchecking the "Allow animation when buffering" checkbox but again no success.
:(

Ok, now I think I get it ... I forgot that the interface was used by others players which don't update it each time it's modified. I should have kept the legacy functions, I will try fixing that today.

[pankov]

Underground78,
thank you very much!
:)

[JohnLai]

Many of use ATI, i cannot reproduce the probleon my ATI.

Can you try this build? http://www.mediafire.com/download.php?9ap6i7kizkwj0a7

Still crashed with StackHash_0a9e. Apparently users Qaq and TheElix also encountered the same problem.
http://forum.doom9.org/showthread.php?t=161047&page=8

[Uxorious]

If it doesn't happen with all computers, it will make the problem really difficult to understand ... :(

You should probably open a ticket on Trac and upload a small 7.1 AAC sample.

Well it happens on all computers I have tried it with.

Just hard to believe I'm the only one throwing 7.1 AAC content at it, but I guess I'll file a Trac issue and see.

Thanks.

[Underground78]

Well it happens on all computers I have tried it with.

This part was not for you, sorry for the misunderstanding.

Just hard to believe I'm the only one throwing 7.1 AAC content at it, but I guess I'll file a Trac issue and see.

Bugs reported here are often lost after a few days that's why we ask everybody to open tickets.

[Uxorious]

Well it happens on all computers I have tried it with.

Just hard to believe I'm the only one throwing 7.1 AAC content at it, but I guess I'll file a Trac issue and see.

Thanks.

Actually I just spotted something.
When I create an AAC file with 7.1 content and play it with MPC-HC, now if I hover my mouse over the FFA icon I see this:
ffdshow audio decoder:
Input: AAC, 6.1, 48000 Hz, 39 kbps (libfaad2)
Output: PCM, 6.1, 48000 Hz, 16-bit integer

I'm guessing it might be a libfaad2 bug...

[Underground78]

@pankov: See the Trac ticket you opened, I have put a test build for you there.

@Uxorious: It might be, I suppose you were using the internal filter at first ?

[v0lt]

Ticket #1606. Do not change the order of the audio tracks. (http://sourceforge.net/apps/trac/mpc-hc/ticket/1606)
What do you think about this?

[sneaker_ger]

Resorting tracks is a bad idea IMHO, because then the order would be different to other programs. People often take a short glimpse at a file with MPC-HC prior to remux/recode to find out the order of the tracks.

[Ayukawa]

Sorry for the newb question, but I'm new to using MPC-HC and using DXVA. Is there any advantage to using DXVA over software decoding if your CPU can handle the decoding? Both in terms of video quality and power consumption?

[Barlow]

Unless you're comparing to a broken decoder video output from the decoder is the same, so quality of the DXVA decoder is not better. However DXVA restricts which video renderer you can use. You won't be able to use madVR, which uses high quality scaling.
Power consumption should be better when using DXVA, as a special purpose piece of hardware will always be more efficient than a general purpose cpu.

[VipZ]

Sorry for the newb question, but I'm new to using MPC-HC and using DXVA. Is there any advantage to using DXVA over software decoding if your CPU can handle the decoding? Both in terms of video quality and power consumption?

I think its quite common that people think DXVA is more efficient, in every test I have run CPU is a fair bit more efficient.

Current total system:
Idle: 325w
CPU (RGB32 HQ): 350-355w avg
CPU (NV12): 355-360w avg
DXVA: 360-365w avg

CPU decoding does fluctuate more than DXVA depending on scene.

In the case of ATI HD5850 paired with a Core i7 860 or i7 2600k, the CPU is more efficient in pretty much every test I have run.

Any recent quad core can play 720p and even some 1080p in mostly idle states.

So DXVA makes for worse efficiency and more complex playback IMO

[SamuriHL]

Can I throw a wrinkle in that discussion? :D LAV CUVID offloads the decoding to the GPU and supports hardware deinterlacing. For a lot of people that's ideal because it allows you to use any renderer you want including madVR. And is great for older CPU's that may not have the power to handle 1080p decoding plus the other tasks necessary for post processing, etc. Just something to consider. :)