Latest OS tools (ffmpeg, mplayer, tsmuxer, ...) discussion

[SubJunk]

Considering that I have it on my computer and going through tests it's certainly more than vapourware

[chocolateboy]

Interesting...

[mazey]

since i dont use subtitles i been trying to get the newer then 32735 mencoder builds to work but i keep getting no sound. does pms code have to be altered in someway in the future if we use the newer ones, or is it yet another new bug introduced.

I'm waiting until VOBSUB support is added before seeing what is causing no sound, so I don't know.

An upcoming SubJunk Build will add support for another project called MPlayer2 which will - if all goes well - replace MPlayer/MEncoder in PMS.
The short story is that the reason MPlayer has kept getting worse for over a year is because the best devs left to make MPlayer2.

sounds like a plan, yes something definelly needs to change it might be another year of waiting for all we know, id be glad to test any replacement and report back..

[ExSport]

@SubJunk:

Dropped features
MEncoder is no longer available
The MEncoder codebase was thoroughly rotten and has been deleted. A different solution to provide some encoding functionality will be added in a future version.

Do you want to say Mplayer2 with encoding support is already usable? Because what I see:

What works
Encoding at variable frame rate (default)
Encoding at constant frame rate using -ofps framerate -oharddup
2-pass encoding (specify flags=+pass1 in the first pass's -ovcopts, specify flags=+pass2 in the second pass)
Hardcoding subtitles using vobsub, ass or srt subtitle rendering (just configure mplayer for the subtitles as usual)
Hardcoding any other mplayer OSD (e.g. time codes, using -osdlevel 3 and -vf expand=::::1)
Encoding directly from a DVD, network stream, webcam, or any other source mplayer supports
Using x264 presets/tunings/profiles (either using @profile in the -ovcopts, or using -vpre profile)
Deinterlacing/Inverse Telecine with any of mplayer's filters for that
Audio file converting: mplayer -o outfile.mp3 infile.flac -novideo -oac libmp3lame -oacopts ab=320k
inverse telecine filters (confirmed working: detc, pullup, filmdint)

What does not work yet
3-pass encoding (ensuring constant total size and bitrate constraints while having VBR audio; mencoder calls this "frameno")
Direct stream copy (is this possible using mplayer's support for video acceleration like VDPAU?)

mostly used feature, "Direct stream copy" doesn't work here

[Madoka]

An upcoming SubJunk Build will add support for another project called MPlayer2 which will - if all goes well - replace MPlayer/MEncoder in PMS.

I was curious as to how close this is to fruition. Or is a lot of work still required?

[Raptor399]

I don't know about you guys, but compiling the tools on OSX is slowly driving me crazy!

The only thing keeping an OSX version of PMS from being released is an update of the OS specific tools.
With an OSX build script in place, building recent versions of all tools is not the problem.
The problem lies in the skewed and distorted sound that mencoder produces.

So I dug into the debug.log to see what is going on. Since PMS runs mencoder with quite a few "-quiet" flags (even a "-really-quiet" was deemed necessary ), I took the relevant commands from the debug.log and made a script to test them:

Code: Select all

#!/bin/bash
# test.sh - Script to test mencoder and tsMuxeR  on OSX without the noisy PMS log messages
# - Install PS3 Media Server
# - Download a QuickTime movie (e.g. from trailers.apple.com), save it as "test.mov"
# - Run this script with "sh test.sh"
MENCODER="/Applications/PS3 Media Server.app/Contents/Resources/Java/osx/mencoder"
TSMUXER="/Applications/PS3 Media Server.app/Contents/Resources/Java/osx/tsMuxeR"

# PMS would create this file
if [ ! -f tsMuxeR.meta ]; then
    cat > tsMuxeR.meta <<EOM
MUXOPT --no-pcr-on-video-pid --new-audio-pes --no-asyncio  --vbr --vbv-len=500
V_MPEG4/ISO/AVC, "video", fps=25, level=4.1, insertSEI, contSPS, track=1
A_AC3, "audio", track=2
EOM
fi

# Same as PMS calls, but stripped of "-quiet" and "-really-quiet" arguments and used simple filenames
"$MENCODER" -ss 0 test.mov -msglevel statusline=2 -ovc copy -nosound -mc 0 -noskip -of rawvideo -o video
"$MENCODER" -ss 0 test.mov -msglevel statusline=2 -channels 6 -ovc copy -of rawaudio -mc 0 -noskip \
    -oac lavc -lavcopts acodec=ac3:abitrate=256 -af lavcresample=48000 -srate 48000 -o audio

"$TSMUXER" tsMuxeR.meta video.mp4


After downloading a nice trailer and saving it as test.mov, running the script produced:

Code: Select all

MEncoder SVN-r33342-4.2.1 (C) 2000-2011 MPlayer Team
success: format: 0  data: 0x0 - 0xc5e222f
libavformat file format detected.
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x1803e00] max_analyze_duration reached
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (aac), -aid 0, -alang eng
VIDEO:  [H264]  1920x816  24bpp  23.976 fps  10188.7 kbps (1243.7 kbyte/s)
[V] filefmt:44  fourcc:0x34363248  size:1920x816  fps:23.976  ftime:=0.0417
videocodec: framecopy (1920x816 24bpp fourcc=34363248)

Video stream: 10188.687 kbit/s  (1273585 B/s)  size: 205411929 bytes  161.286 secs  3867 frames
MEncoder SVN-r33342-4.2.1 (C) 2000-2011 MPlayer Team
success: format: 0  data: 0x0 - 0xc5e222f
libavformat file format detected.
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x1803e00] max_analyze_duration reached
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (aac), -aid 0, -alang eng
VIDEO:  [H264]  1920x816  24bpp  23.976 fps  10188.7 kbps (1243.7 kbyte/s)
[V] filefmt:44  fourcc:0x34363248  size:1920x816  fps:23.976  ftime:=0.0417
==========================================================================
Opening audio decoder: [ffmpeg] FFmpeg/libavcodec audio decoders
AUDIO: 44100 Hz, 2 ch, s16le, 98.2 kbit/6.96% (ratio: 12272->176400)
Selected audio codec: [ffaac] afm: ffmpeg (FFmpeg AAC (MPEG-2/MPEG-4 Audio))
==========================================================================
Ignoring video stream!
videocodec: framecopy (1920x816 24bpp fourcc=34363248)
Encoder requires input not properly supported.Try using e.g. 'ac3_fixed' instead of 'ac3' as codec
[ac3 @ 0xd9faa0]channel_layout not specified
[ac3 @ 0xd9faa0]No channel layout specified. The encoder will guess the layout, but it might be incorrect.

Video stream: 10188.687 kbit/s  (1273585 B/s)  size: 205411929 bytes  161.286 secs  3867 frames

Audio stream:  256.000 kbit/s  (32000 B/s)  size: 5177344 bytes  161.792 secs
SmartLabs tsMuxeR.  Version 1.10.6  http://www.smlabs.net
Assertion failed: ((*buffer & 0x80) == 0), function deserialize, file ../vodTransport/src/nalUnits.cpp, line 210.
test.sh: line 18: 67847 Abort trap              "$TSMUXER" tsMuxeR.meta video.mp4


Those last lines are probably my fault, as I have no idea how tsMuxeR works.
But look at the audio output. Mencoder seems to be complaining about "ac3" as codec.

To make things interesting, here is the output of the same script and movie, but this time with the old (working) mencoder:

Code: Select all

MEncoder SVN-r29241-snapshot-4.0.1 (C) 2000-2009 MPlayer Team
success: format: 0  data: 0x0 - 0xc5e222f
libavformat file format detected.
[lavf] Video stream found, -vid 0
[lavf] Audio stream found, -aid 1
VIDEO:  [avc1]  1920x816  24bpp  23.976 fps    0.0 kbps ( 0.0 kbyte/s)
[V] filefmt:44  fourcc:0x31637661  size:1920x816  fps:23.976  ftime:=0.0417
videocodec: framecopy (1920x816 24bpp fourcc=31637661)

Video stream: 10188.697 kbit/s  (1273587 B/s)  size: 205411929 bytes  161.286 secs  3867 frames
MEncoder SVN-r29241-snapshot-4.0.1 (C) 2000-2009 MPlayer Team
success: format: 0  data: 0x0 - 0xc5e222f
libavformat file format detected.
[lavf] Video stream found, -vid 0
[lavf] Audio stream found, -aid 1
VIDEO:  [avc1]  1920x816  24bpp  23.976 fps    0.0 kbps ( 0.0 kbyte/s)
[V] filefmt:44  fourcc:0x31637661  size:1920x816  fps:23.976  ftime:=0.0417
==========================================================================
Opening audio decoder: [faad] AAC (MPEG2/4 Advanced Audio Coding)
FAAD: compressed input bitrate missing, assuming 128kbit/s!
AUDIO: 44100 Hz, 2 ch, s16le, 128.0 kbit/9.07% (ratio: 16000->176400)
Selected audio codec: [faad] afm: faad (FAAD AAC (MPEG-2/MPEG-4 Audio))
==========================================================================
Ignoring video stream!
videocodec: framecopy (1920x816 24bpp fourcc=31637661)
[ac3 @ 0x104c400]No channel layout specified. The encoder will guess the layout, but it might be incorrect.

Video stream: 10188.697 kbit/s  (1273587 B/s)  size: 205411929 bytes  161.286 secs  3867 frames

Audio stream:  256.000 kbit/s  (31999 B/s)  size: 5161984 bytes  161.312 secs
SmartLabs tsMuxeR.  Version 1.10.6  http://www.smlabs.net
Assertion failed: ((*buffer & 0x80) == 0), function deserialize, file ../vodTransport/src/nalUnits.cpp, line 210.
test.sh: line 18: 67866 Abort trap              "$TSMUXER" tsMuxeR.meta video.mp4


The old (working) mencoder chooses a different audio codec and it does not complain about using ac3.

I see that in r575, PMS got fresh win32 tools. But they are Sherpya builds, specific for win32. From what I can see, the versions are:

mencoder: MEncoder Sherpya-SVN-r30369-4.2.5
mencoder_mt: MEncoder Sherpya-SVN-r30369-4.2.5
mencoder-alternate: MEncoder Sherpya-SVN-r32674-4.2.5
mencoder-alternate_mt: MEncoder Sherpya-MT-SVN-r32674-4.2.5

I tried to build those revisions of MPlayer, but they don't seem to compile for me on OSX. I would not mind helping out, but I must admit I have no clue on how to move forward on this and get the OS tools back in working shape on OSX.
Sherpya did post his diffs, but they seem specific for win32 and for his setup. No clue if they would work for OSX. Besides, I can imagine that adding a dependency to Sherpya's diffs on top of the rapidly changing MPlayer source only causes more uncertainty.

Sooo... Anyone with bright ideas?

[Raptor399]

Looking in detail at the diffs of the Sherpya release makes me wonder why that release was chosen.
It looks like a lot of stuff is added or changed that PMS does not need at all.

Which diffs in particular make the Sherpya releases better for PMS than a regular mplayer build?

[SubJunk]

It's just the most popular Windows build. I don't see that there's anything to be gained by building myself, and I guess Shagrath felt the same way

[mazey]

It's just the most popular Windows build. I don't see that there's anything to be gained by building myself, and I guess Shagrath felt the same way

how far away is mplayer2 from being usable, i dont really know just by looking at there page, thats why im asking heh

[SubJunk]

I haven't tested it yet because the windows build generates a false-positive with my AV program Avira, and since that's a popular AV program we can't include mplayer2 until that problem is resolved.
Unfortunately Avira don't seem very enthusiastic about fixing it.