Media Player Classic - BE Win32/x64

[Aleksoid1978]

try after maximize, I can reproduce that way.
Ok - i see. Thanks.

P.S. Fixed.

[Aleksoid1978]

Thanks!
With version above, videos are fluid. I feel real 60fps watching videos. With previous version, the motion was stuttered like fps was cut in a half.
But at the same time, deinterlacing quality is worse. Its like different deinterlacing method is used? I cannot distinguish between dx9 vs dx11. But when installing previous version (for expample MPCVideoRenderer-v0.3.3.907) deinterlacing quality is superb with dx11 but worse with dx9.
Example on test videos:
bad https://imgur.com/BNVKLoG
good https://imgur.com/9dAfm8W
Actually this is what i consider watching test patterns from this site https://www.uydudoktoru.com/threads/hd-1080i-test-pattern-detect-vector-adaptive-deinterlacing-others.6689/
Edit
I hope i was not wrong. Its just for me lines are smoother in previous versions, but 1P area should not be black&white but with stripes, so i dont know.
Eventually this version may stay as is, as MPC renderer gives same results as evr.

Can you check/test DX11 deinterlace in other player, as example in standart Windows 10 player or VLC using DX11 output ?

[Liisachan]

Once it's maximized, the results of [Alt]+[1]/[2]/[3] are all strange, both from full-screened or windowed.

Also, if you're using a keyboard layout where the right Alt key is AltGr and you hit e.g. Right [Alt]+[2], the result is unexpected and potentially confusing: the same as if you just hit [2] (Presets | Compact). The Russian keyboard is one example of AltGr layouts, if Right Alt + 8 is ₽.

If you're using a keyboard layout where the right Alt works like the left Alt, then Right [Alt]+[1]/[2]/[3] are same as Left [Alt] + Number. Maybe not important. Just something I've noticed while testing [Alt]+[2].

[huhn]

it's hard to test if deinterlacing works fine with mpcVR because control+ arrow key moves the image not by deinterlaced fields but by frame.

BTW. on AMD i haven't notice anything that is working differently between DXVA2 deinterlacing and d3d11.

[v0lt]

@huhn
D3D11 deinterlacing does not work on AMD cards.
We are now checking D3D11 deinterlacing in VLC 3.0.8, it seems it does not work there either.
Perhaps a driver bug.

[Aleksoid1978]

https://yadi.sk/i/mHJfXwFmFLzHug
Good sample for test deinterlace working or not.

[huhn]

it's clearly a driver bug and working at the same time the quality is just very low i guess a simple bob deinterlacing is done here but it is still working.

interesting sample it doesn't work properly with amd for again both DXVA2 and D3D11.

there are currently 3 major bugs in DXVA2 processing for AMD:
range levels.
deinterlacing is of low quality it.
DXVA native decoding of 10 bit sources produces a lot of banding.

only the deinterlacing bug is the same in d3d11.

[littleD]

https://yadi.sk/i/mHJfXwFmFLzHug
Good sample for test deinterlace working or not.@Aleksoid1978
Video sample You posted is fluid 60fps on intel. I think MPCVideoRenderer-v0.3.3.912_git2019.11.01-a6c2592 version is okay! Deinterlacing works of course.

Sorry for the previous confusion about deinterlacing quality. It is just Intel Hardware deinterlacing does not expect my needs. It is using only Adaptive which is also not best in quality. AMD and Nvidia use some sort of better algorithm Vector Adaptive.

VLC uses software deinterlacing, no? And its also not 100% proper i checked yesterday. Laf filters beats it, also in software.

So far, almost ideal results give me lavfilters, copyback (with dx11 choose hardware specifically), and software Yadif 60FPS. When watching test pattern, 1P area is filled with stripes as it should. And all sloped lines are antialiased. I am talking about quality now. Intel hardware is worse than software yadif in lav filters (vlc quality is also worse than laf filters) but that should be no interest of mpc devs. This is different area. MPC vrenderer is doing deinterlacing and video output is fluid on intel.

[littleD]

it's clearly a driver bug and working at the same time the quality is just very low i guess a simple bob deinterlacing is done here but it is still working.

interesting sample it doesn't work properly with amd for again both DXVA2 and D3D11.

there are currently 3 major bugs in DXVA2 processing for AMD:
range levels.
deinterlacing is of low quality it.
DXVA native decoding of 10 bit sources produces a lot of banding.

only the deinterlacing bug is the same in d3d11.
It cant be such bad, that amd drivers do not deinterlace at all? Or doing bob only. Maybe wait for implementing deinterlacing by the team.
Also Range level should be fixed yeras ago (i had amd previously).
DXVA2 10 bit on intel is good, but HDR to sdr conversion is bad. With dx11 is ok.

@huhn
During playback interlaced stream, go to properties of MPC v renderer and click on Information Tab. It gives Deinterlacetechnology available. When there is more than one, driver should choose best quality algo automatically. If not, then there is bug indeed.

[huhn]

https://abload.de/img/untitled81k35.jpg

the test version of mpcVR doesn't work at all that DXVA2. so your sample may work fine with older versions.

edit: v0.3.2.860:
AMD deinterlace:
d3d11 works in low quality no info in information screen about deinterlacing. let me guess it is just discarding half the fields?
dxva2 broken doing nothing info shows pixeladaptive.
your test file works too.
Also Range level should be fixed yeras ago (i had amd previously).
trust me this is not the first time that the best deinterlacing AMD was offering was terrible the same happend when polaris was released and they create a new driver UI where up to this day you can't even switch the refreshrate.

it's clearly a driver bug and working at the same time the quality is just very low i guess a simple bob deinterlacing is done here but it is still working.
is bob on every video renderer i tested.

[littleD]

Yea, i believe you. Amd was doing bad things to its drivers in the past.
but Aleksoid1978 said mpc vr is not working with AMD yet.
Also DDU remove drivers and fresh install helped on many problems (also, registry problems) in the past.
EDIT: ah yes, what if You have to use yv12 instead of nv12??

@Huhn, lets open DXVA checker and go to Processordevice>>dxva1/2 and move down to Deinterlave Technology. Do You see YES at Motion Vector steered? I ithink common description for this algo is Vector Adaptive (best available on hardware). If yes, you may try with evr renderer too.

@aleksoid1978
Is that possible to show actual deinterlacing algo in use? In MPC properties i see all five. I just wonder, if Motion compensate isnt better than Adaptive algo. But is deinterlacing algorithm what player/renderer can choose on windows machine at all?

MS docs knowledge:
dxva2 https://docs.microsoft.com/pl-pl/windows-hardware/drivers/ddi/dxva/ne-dxva-_dxva_deinterlacetech
dx11 https://docs.microsoft.com/en-us/windows/win32/api/d3d11/ne-d3d11-d3d11_video_processor_processor_caps

[Aleksoid1978]

https://abload.de/img/untitled81k35.jpg

the test version of mpcVR doesn't work at all that DXVA2. so your sample may work fine with older versions.

edit: v0.3.2.860:
AMD deinterlace:
d3d11 works in low quality no info in information screen about deinterlacing. let me guess it is just discarding half the fields?
dxva2 broken doing nothing info shows pixeladaptive.
your test file works too.

trust me this is not the first time that the best deinterlacing AMD was offering was terrible the same happend when polaris was released and they create a new driver UI where up to this day you can't even switch the refreshrate.


is bob on every video renderer i tested.

Upload this video for test.

[huhn]

if DXVA checker is correct it needs NV12 for DXVA2 deinterlacing nothing else is listed.

and i tested different renderer already. i wouldn't come to the conclusion that AMD driver are broken if i didn't even do that.

DXVA2 only list pixel adaptive for vector adaptive deint.
for softwaredevice it only list bob...

d3d11 has everything except blend.

[huhn]

Upload this video for test.

it's 3:2 telecine i prefer this for deint tests. but sure i can make you a sample.

[Aleksoid1978]

Yea, i believe you. Amd was doing bad things to its drivers in the past.
but Aleksoid1978 said mpc vr is not working with AMD yet.
Also DDU remove drivers and fresh install helped on many problems (also, registry problems) in the past.
EDIT: ah yes, what if You have to use yv12 instead of nv12??

@Huhn, lets open DXVA checker and go to Processordevice>>dxva1/2 and move down to Deinterlave Technology. Do You see YES at Motion Vector steered? I ithink common description for this algo is Vector Adaptive (best available on hardware). If yes, you may try with evr renderer too.

@aleksoid1978
Is that possible to show actual deinterlacing algo in use? In MPC properties i see all five. I just wonder, if Motion compensate isnt better than Adaptive algo. But is deinterlacing algorithm what player/renderer can choose on windows machine at all?

MS docs knowledge:
dxva2 https://docs.microsoft.com/pl-pl/windows-hardware/drivers/ddi/dxva/ne-dxva-_dxva_deinterlacetech
dx11 https://docs.microsoft.com/en-us/windows/win32/api/d3d11/ne-d3d11-d3d11_video_processor_processor_caps

For DX11 we can select only processor device but not a deinterlace technology, in properties you see what it support.

[huhn]

sample:
http://s000.tinyupload.com/index.php?file_id=90576381487986185687

just as a reminder and i just retested it:
DXVA2:
v0.3.2.860: broken.
current test: broken
d3d11:
v0.3.2.860: "fine"
current test: broken

using an AMD navi card.

For DX11 we can select only processor device but not a deinterlace technology, in properties you see what it support.

so no way to ask for inverse telecine?

[littleD]

For DX11 we can select only processor device but not a deinterlace technology, in properties you see what it support.OKay. I dig out my laptop with old ati card. I can test too.

@huhn
i cannot decode this video too without jaggies. VLC cannot decode IVTC. Are you sure remuxing it to mkv was okay?

EVR CP do some IVTC

[huhn]

look at the screen it is combing the fields or with other word it is doing nothing not even bobbing.

nvidia works on it to and madVR IVTC works fine too but that doesn't matter here.

here is what it should look like when deint actually works and the current quality of AMD deint.
https://abload.de/img/amddeint5njc0.jpg

the image is not the same frame and both a jpg so no point in comparing them.

every telecine file can be deinterlaced without a question the field order is correct so yes makemkv didn't screw up that's the only thing that matter for deint.

[littleD]

look at the screen it is combing the fields or with other word it is doing nothing not even bobbing.

nvidia works on it to and madVR IVTC works fine too but that doesn't matter here.

here is what it should look like when deint actually works and the current quality of AMD deint.
https://abload.de/img/amddeint5njc0.jpg

the image is not the same frame and both a jpg so no point in comparing them.

every telecine file can be deinterlaced without a question the field order is correct so yes makemkv didn't screw up that's the only thing that matter for deint.I run some tests on laptop with 5xxx series :)

Dxva2 deinterlace is ok on this old machine, Videos 29,97i is fluid. I see also pixel adaptive. Your sample decodes like on your last picture.
I cannot test dx11 because video does not start, i have win 8.1 and old drivers. Anyway i believe mpc vr should work at least in dxva2. I think something is really wrong with drivers.

[huhn]

driver 13.12 should be fine. it's kind of an legendary driver for madVR.
and yes i know it is 6 years only.

this driver should have an option to select the deint type.

[Aleksoid1978]

sample:
http://s000.tinyupload.com/index.php?file_id=90576381487986185687

Test on Nvidia 960 - perfect deinterlace in EVR/EVR-CP/MPC VR(DX9/DX11) :)

[littleD]

driver 13.12 should be fine. it's kind of an legendary driver for madVR.
and yes i know it is 6 years only.

this driver should have an option to select the deint type.Yea, there were nice old times with amd drivers too.. And yea i can select deint method

Regarding 3:2, did you eneble pulldown in drivers? I remember older dxvachecker versions were able to change amd registry settings. Decoders to use and pulldown too maybe. I think the might be some issue.

[huhn]

everything is deint in an absolute pathetic quality. true interlaced content is flickering is just broken.

for your own sanity is better if you don't look up what the "option" are in this "modern" driver.

[s0meone_new]

It would be necessary to improve the Hungarian translation. Translation is incomplete. I can help the maker if he wants this.
Send me a private message.

[Klaus1189]

@s0meone_new:
Download the txt file here:
https://sourceforge.net/p/mpcbe/code/4877/tree/trunk/src/apps/mpcresources/text/

Replace the „4877“ in the link with the current version in the future, you check for the current version here:
https://sourceforge.net/p/mpcbe/code/commit_browser

Then edit the file and upload it to Google drive or similar and post link ;)

[s0meone_new]

Klaus1189:
Okay! :)
I prepare it after some days.

[v0lt]

The current version of translations (HEAD) can be taken here (https://sourceforge.net/p/mpcbe/code/HEAD/tree/trunk/src/apps/mpcresources/text/).

@beter
Thank. Updated in r4874.

[Klaus1189]

s0meone_new
If you want to change something in the installer, download the following file and edit the parts of your language:
https://sourceforge.net/p/mpcbe/code/4877/tree/trunk/distrib/custom_messages.iss

When done upload the file again to Google drive or similar and post link here ;)

[s0meone_new]

v0lt, Klaus1189:
custom_messages.iss completed:
https://www12.zippyshare.com/v/sH0nFBUM/file.html

some tips:
1.
This is a nice file icon. Should be put in the installer.
FlatroBEx64_Icons_v1.0.rar
https://sourceforge.net/projects/mpcbe/files/Icons/
2.
default setting in the player: Options-Keys: Play/Pause - Mouse Windowed and Mouse Fullscreen it is on. Why? Very disturbing.

[Liisachan]

4877 tested:
Maximize -> [Alt]+[2] is ok now!
But Maximize -> [Alt]+[1] / [Alt]+[3] doesn't work properly for me: the results are same as [Alt]+[2]

[s0meone_new]

mplayerc.hu.rc.txt translation in progress, but rather incomplete. It takes several days for me to finish.

[Aleksoid1978]

4877 tested:
Maximize -> [Alt]+[2] is ok now!
But Maximize -> [Alt]+[1] / [Alt]+[3] doesn't work properly for me: the results are same as [Alt]+[2]

Thanks :) Fixed.

[v0lt]

1.
This is a nice file icon. Should be put in the installer.
FlatroBEx64_Icons_v1.0.rar
We do not add such things to the installer.
2.
default setting in the player: Options-Keys: Play/Pause - Mouse Windowed and Mouse Fullscreen it is on. Why? Very disturbing.This is standard behavior for MPC-BE, MPC-HC and other players.

Thank. custom_messages.iss updated in r4879.

[s0meone_new]

mplayerc.hu.rc.txt: https://www106.zippyshare.com/v/RCczRzIs/file.html
Translation is incomplete not 100%, but I translated important texts. If it seems necessary to continue, then i will continue. :)

the name: "s0meone_new" I only use it here. :) I use that name: mickey

[Liisachan]

Thanks :) Fixed.

Thanks! Tested 4878. Working fine.

@s0meone_new
(aside) Gabest, the very first person who created MPC (from which MPC-HC and MPC-BE were born), is from Hungary too iirc.

[Klaus1189]

I think there is a mistake in line 76 of custom_messages.iss in r4879:
hu.WelcomeLabel2 shouldn't be translated to hu.ÜdvözöljükLabel2, otherwise it won't work anymore.
https://sourceforge.net/p/mpcbe/code/4879/tree//trunk/distrib/custom_messages.iss?diff=54bb00b7c4d104415b75022c:4878

[v0lt]

@s0meone_new
Thank. Updated in r4880.

@Klaus1189
Thank. Fixed in r4882.

[s0meone_new]

Klaus1189:
Sorry. I only change the words. :) I didn't change another thing.

Üdvözöljük = Welcome

[Klaus1189]

Don't worry ;) The words left from "=" must not (are not allowed to) be translated, because these are the names of the GUI labels and the text right from the "=" should be translated, because these words will be displayed on the label with the name hu.WelcomeLabel2 is this case.

[s0meone_new]

Yes i already know. :D
mplayerc.hu.rc.tx is that okay? I can fix it if there is a problem somewhere.

I continue to translate! :)

[Klaus1189]

I cannot find a problem of syntax of the file, but for grammar mistakes I can not help because I don’t understand that language :)
Please check the new translation when a nightly build is available if every new string is correct and no typo is in the translation. I must also say, well done many translated strings, I know how time consuming that is but MPC-BE is worth every minute I spent for the translation. Coded with passion for perfection by the devs ;)

[s0meone_new]

Yes, yes and YES! :D
Tomorrow it will be completely ready.
Cool? :)

Please check it out: https://www105.zippyshare.com/v/KgEfERMY/file.html
90% it will be complete tomorrow. :)

[Liisachan]

When MP4 has 2 or more cover arts, [Shift]+[F10] -> resources shows only one (the first?) embedded file, if the internal MP4/MOV source filter is enabled in MPC-BE. If the internal filter is disabled (MPC-BE + LAV) or MPC-HC is used, you can see all embedded files via [Shift]+[F10].

Not a big problem at all; happened to notice.

Sample MP4 (55 KiB) (http://faireal.net/image/2019/two-covers.mp4) with 2 cover arts (PNG, Jpeg)

[s0meone_new]

Hungarian translation finished! :)
https://www101.zippyshare.com/v/pSDLUE88/file.html

[Aleksoid1978]

When MP4 has 2 or more cover arts, [Shift]+[F10] -> resources shows only one (the first?) embedded file, if the internal MP4/MOV source filter is enabled in MPC-BE. If the internal filter is disabled (MPC-BE + LAV) or MPC-HC is used, you can see all embedded files via [Shift]+[F10].

Not a big problem at all; happened to notice.

Sample MP4 (55 KiB) (http://faireal.net/image/2019/two-covers.mp4) with 2 cover arts (PNG, Jpeg)

Thanks - fixed :)

[v0lt]

@s0meone_new
Thank. Updated in r4885.

[s0meone_new]

where is the r4885 installer available? I can't verify it.

[ryrynz]

where is the r4885 installer available? I can't verify it.

Wait, they will update. Beta links a page or two back.

[Liisachan]

Thanks - fixed :)

Tested 4885 with a sample MP4 (http://faireal.net/image/2019/3covers.mp4) (no video, silent audio + 3 cover arts in BMP/PNG/JPG). Everything works fine! Thanks :D

But I have 2 questions:
1) dataType 0x1b (for BMP) seems undocumented. do we really have to support non-standard BMP cover arts? It won't hurt though...
2) does the GETU32(db->GetData()) check in MP4Splitter.cpp necessary? I'm guessing something like this might be simpler and quicker:

if (dataType == 0x0d) { /* jpeg */ }
else if (dataType == 0x0e) { /* png */ }
else if (dataType == 0x1b) { /* bmp */ } // if to be supported at all

[Aleksoid1978]

But I have 2 questions:
1) dataType 0x1b (for BMP) seems undocumented. do we really have to support non-standard BMP cover arts? It won't hurt though...
2) does the GETU32(db->GetData()) check in MP4Splitter.cpp necessary? I'm guessing something like this might be simpler and quicker:

if (dataType == 0x0d) { /* jpeg */ }
else if (dataType == 0x0e) { /* png */ }
else if (dataType == 0x1b) { /* bmp */ } // if to be supported at all


Why it for you ? :)