H.264 Audio/Video out of sync

[magicite]

I too am having slight audio-video sync issues. Approaching half a second in a 2 hour long movie. Using latest version of ffmpegX (t r3), source is a DVD ('Constantine') ripped using MTR. Had the exact same problem with the original t.

[lhawkins]

OK, I tried "Fight Club" again and it is still out of sync by about 3/4 second by the end... Here is the output from the command line. I used auto crop so the output was set to 640x272 2.35:1 at NTSC FILM frame rate. Audio was 96khz AAC, filters at default settings and use CABAC, B-Frames, VBR Constant Q=20 and multi-hexagon Motion Estimation. Oh, 2 pass encoding as well! (I've had the same issues with a one pass encode though) The quality of the video is excellent, but with the sync issues it is useless. The encode was done on a G5/1.6 and it took two days (This was because other processor intensive apps were running in the background as well, usually this type of encode takes around 6-8 hours on this machine)

COMMAND LINE OUTPUT
-----------------------------
Laurence-Hawkins:~ lhawkins$ /Applications/ffmpegX.app//Contents/Resources/pbpaste | sh
Encoding started on Mon Sep 12 12:41:04 PDT 2005
MEncoder dev-CVS-050814-13:46-3.3 (C) 2000-2005 MPlayer Team
Mac OSX static build for ffmpegX
AltiVec found
CPU: PowerPC
File not found: 'frameno.avi'
Failed to open frameno.avi
libdvdread: Couldn't find device name.
Reading disc structure, please wait...
There are 6 titles on this DVD.
There are 37 chapters in this DVD title.
There are 1 angles in this DVD title.
DVD successfully opened.
success: format: 0 data: 0x0 - 0xd8cbb800
MPEG-PS file format detected.
VIDEO: MPEG2 720x480 (aspect 3) 29.970 fps 9800.0 kbps (1225.0 kbyte/s)
[V] filefmt:2 fourcc:0x10000002 size:720x480 fps:29.97 ftime:=0.0334
Opening video filter: [expand osd=1]
Expand: -1 x -1, -1 ; -1, osd: 1, aspect: 0.000000, round: 1
Opening video filter: [scale w=640 h=272]
Opening video filter: [crop w=716 h=358 x=2 y=58]
Crop: 716 x 358, 2 ; 58
Opening video filter: [pp=0x33]
Opening video filter: [harddup]
================================================== ========================
Opening video decoder: [mpegpes] MPEG 1/2 Video passthrough
VDec: vo config request - 720 x 480 (preferred csp: Mpeg PES)
[PP] Using external postprocessing filter, max q = 6.
Could not find matching colorspace - retrying with -vf scale...
Opening video filter: [scale]
The selected video_out device is incompatible with this codec.
Also try adding the scale filter, e.g. -vf spp,scale instead of -vf spp.
VDecoder init failed
Opening video decoder: [libmpeg2] MPEG 1/2 Video decoder libmpeg2-v0.4.0b
Selected video codec: [mpeg12] vfm:libmpeg2 (MPEG-1 or 2 (libmpeg2))
================================================== ========================
Writing AVI header...
VDec: vo config request - 720 x 480 (preferred csp: Planar YV12)
[PP] Using external postprocessing filter, max q = 6.
VDec: using Planar YV12 as output csp (no 0)
Movie-Aspect is 1.78:1 - prescaling to correct movie aspect.
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2

SwScaler: FAST_BILINEAR scaler, from Planar YV12 to Planar YV12 using AltiVec
x264 [info]: using SAR=129/128
x264 [info]: using cpu capabilities Altivec
Pos: 0.0s 2f ( 0%) 0fps Trem: 0min 0mb A-V:0.000 [0:0]
1 duplicate frame(s)!
Pos: 0.1s 4f ( 0%) 2fps Trem: 0min 0mb A-V:0.000 [0:0]
1 duplicate frame(s)!
Pos: 0.4s 11f ( 0%) 2fps Trem: 0min 0mb A-V:0.000 [0:0]
demux_mpg: 24000/1001fps progressive NTSC content detected, switching framerate.
Pos:8350.4s 200209f (99%) 47fps Trem: 0min 480mb A-V:0.000 [483:0]
Flushing video frames

Writing AVI index...
Fixing AVI header...

Video stream: 483.166 kbit/s (60395 bps) size: 504328242 bytes 8350.392 secs 200209 frames
x264 [info]: slice I:3108 Avg QP:20.31 Avg size: 13802 PSNR Mean Y:45.03 U:47.77 V:49.27 Avg:45.86 Global:45.57
x264 [info]: slice P:100634 Avg QP:20.81 Avg size: 3542 PSNR Mean Y:44.88 U:47.73 V:48.91 Avg:45.70 Global:45.33
x264 [info]: slice B:96467 Avg QP:22.82 Avg size: 1076 PSNR Mean Y:44.37 U:47.57 V:48.73 Avg:45.25 Global:44.87
x264 [info]: slice I Avg I4x4:57.9% I8x8:0.0% I16x16:42.1%
x264 [info]: slice P Avg I4x4:4.3% I8x8:0.0% I16x16:6.8% P:46.6% P8x8:4.9% PSKIP:37.4%
x264 [info]: slice B Avg I4x4:0.7% I8x8:0.0% I16x16:0.8% P:16.5% B:7.5% B8x8:2.4% DIRECT:3.1% BSKIP:69.1%
x264 [info]: PSNR Mean Y:44.64 U:47.65 V:48.83 Avg:45.48 Global:45.11 kb/s:482.1
MEncoder dev-CVS-050814-13:46-3.3 (C) 2000-2005 MPlayer Team
Mac OSX static build for ffmpegX
AltiVec found
CPU: PowerPC
File not found: 'frameno.avi'
Failed to open frameno.avi
libdvdread: Couldn't find device name.
Reading disc structure, please wait...
There are 6 titles on this DVD.
There are 37 chapters in this DVD title.
There are 1 angles in this DVD title.
DVD successfully opened.
success: format: 0 data: 0x0 - 0xd8cbb800
MPEG-PS file format detected.
VIDEO: MPEG2 720x480 (aspect 3) 29.970 fps 9800.0 kbps (1225.0 kbyte/s)
[V] filefmt:2 fourcc:0x10000002 size:720x480 fps:29.97 ftime:=0.0334
Opening video filter: [expand osd=1]
Expand: -1 x -1, -1 ; -1, osd: 1, aspect: 0.000000, round: 1
Opening video filter: [scale w=640 h=272]
Opening video filter: [crop w=716 h=358 x=2 y=58]
Crop: 716 x 358, 2 ; 58
Opening video filter: [pp=0x33]
Opening video filter: [harddup]
================================================== ========================
Opening video decoder: [mpegpes] MPEG 1/2 Video passthrough
VDec: vo config request - 720 x 480 (preferred csp: Mpeg PES)
[PP] Using external postprocessing filter, max q = 6.
Could not find matching colorspace - retrying with -vf scale...
Opening video filter: [scale]
The selected video_out device is incompatible with this codec.
Also try adding the scale filter, e.g. -vf spp,scale instead of -vf spp.
VDecoder init failed
Opening video decoder: [libmpeg2] MPEG 1/2 Video decoder libmpeg2-v0.4.0b
Selected video codec: [mpeg12] vfm:libmpeg2 (MPEG-1 or 2 (libmpeg2))
================================================== ========================
Writing AVI header...
VDec: vo config request - 720 x 480 (preferred csp: Planar YV12)
[PP] Using external postprocessing filter, max q = 6.
VDec: using Planar YV12 as output csp (no 0)
Movie-Aspect is 1.78:1 - prescaling to correct movie aspect.
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2
SwScaler: reducing / aligning filtersize 2 -> 2

SwScaler: FAST_BILINEAR scaler, from Planar YV12 to Planar YV12 using AltiVec
x264 [info]: using SAR=129/128
x264 [info]: using cpu capabilities Altivec
Pos: 0.0s 2f ( 0%) 0fps Trem: 0min 0mb A-V:0.000 [0:0]
1 duplicate frame(s)!
Pos: 0.1s 4f ( 0%) 0fps Trem: 0min 0mb A-V:0.000 [0:0]
1 duplicate frame(s)!
Pos: 0.4s 11f ( 0%) 0fps Trem: 0min 0mb A-V:0.000 [0:0]
demux_mpg: 24000/1001fps progressive NTSC content detected, switching framerate.
Pos:8350.4s 200209f (99%) 47fps Trem: 0min 487mb A-V:0.000 [490:0]
Flushing video frames

Writing AVI index...
Fixing AVI header...

Video stream: 490.052 kbit/s (61256 bps) size: 511516243 bytes 8350.392 secs 200209 frames
x264 [info]: slice I:3108 Avg QP:20.36 Avg size: 13707 PSNR Mean Y:45.00 U:47.75 V:49.25 Avg:45.84 Global:45.53
x264 [info]: slice P:100634 Avg QP:20.57 Avg size: 3649 PSNR Mean Y:45.00 U:47.81 V:49.00 Avg:45.81 Global:45.47
x264 [info]: slice B:96467 Avg QP:22.82 Avg size: 1043 PSNR Mean Y:44.42 U:47.62 V:48.80 Avg:45.30 Global:44.93
x264 [info]: slice I Avg I4x4:57.8% I8x8:0.0% I16x16:42.2%
x264 [info]: slice P Avg I4x4:4.5% I8x8:0.0% I16x16:6.7% P:47.0% P8x8:5.1% PSKIP:36.7%
x264 [info]: slice B Avg I4x4:0.7% I8x8:0.0% I16x16:0.8% P:16.2% B:7.4% B8x8:2.2% DIRECT:3.0% BSKIP:69.6%
x264 [info]: PSNR Mean Y:44.72 U:47.72 V:48.91 Avg:45.56 Global:45.20 kb/s:489.0
MPlayer dev-CVS-050814-13:46-3.3 (C) 2000-2005 MPlayer Team
Mac OSX static build for ffmpegX
AltiVec found
CPU: PowerPC

Playing dvd://1.
libdvdread: Couldn't find device name.
Reading disc structure, please wait...
There are 6 titles on this DVD.
There are 37 chapters in this DVD title.
There are 1 angles in this DVD title.
DVD successfully opened.
MPEG-PS file format detected.
VIDEO: MPEG2 720x480 (aspect 3) 29.970 fps 9800.0 kbps (1225.0 kbyte/s)
================================================== ========================
Opening audio decoder: [liba52] AC3 decoding with liba52
Using AltiVec optimized IMDCT transform
AC3: 2.0 (dolby) 48000 Hz 192.0 kbit/s
Checking for AltiVec resampler : 0x0000000a, 2
Using AltiVec optimized resampler
AUDIO: 48000 Hz, 2 ch, s16be, 192.0 kbit/12.50% (ratio: 24000->192000)
Selected audio codec: [a52] afm:liba52 (AC3-liba52)
================================================== ========================
================================================== ========================
Forced video codec: dummy
Cannot find codec matching selected -vo and video format 0x10000002.
Read DOCS/HTML/en/codecs.html!
================================================== ========================
Building audio filter chain for 48000Hz/2ch/s16be -> 0Hz/0ch/s8...
[AO PCM] File: audiodump.wav (WAVE)
PCM: Samplerate: 48000Hz Channels: Stereo Format s16le
[AO PCM] Info: fastest dumping is achieved with -vc dummy -vo null
PCM: Info: to write WAVE files use -waveheader (default).
AO: [pcm] 48000Hz 2ch s16le (2 bps)
Building audio filter chain for 48000Hz/2ch/s16be -> 48000Hz/2ch/s16le...
Video: no video
Starting playback...
A:8350.0 ( 2:19:10.0) 5.1%

Exiting... (End of file)
ffmpeg version 0.4.9-pre1, build 4736, Copyright (c) 2000-2004 Fabrice Bellard
configuration: --enable-mp3lame --enable-gpl --disable-vhook --enable-ffplay --disable-ffserver --enable-a52 --enable-xvid --enable-faac --enable-faad --enable-amr_nb --enable-amr_wb --enable-pthreads
built on Jan 7 2005 02:34:06, gcc: 3.3 20030304 (Apple Computer, Inc. build 1666)
Input #0, wav, from '/Users/lhawkins/Desktop/FIGHTCLB/20real-2pass-MultiHex/audiodump.wav':
Duration: 02:19:10.3, start: 0.000000, bitrate: 1536 kb/s
Stream #0.0: Audio: pcm_s16le, 48000 Hz, stereo, 1536 kb/s
Output #0, mp4, to '/Users/lhawkins/Desktop/FIGHTCLB/20real-2pass-MultiHex.aac.mp4':
Stream #0.0: Audio: aac, 48000 Hz, stereo, 96 kb/s
Stream mapping:
Stream #0.0 -> #0.0
Press [q] to stop encoding
size= 197378kB time=8350.3 bitrate= 193.6kbits/s
video:0kB audio:194319kB global headers:0kB muxing overhead 1.574006%
AVC-H264 import - frame size 640 x 272 at 23.976 FPS
Adjusting AVC SizeLength to 16 bits
Import results: 200209 samples - Slices: 6216 I 201268 P 192934 B - 1 SEI - 5400 IDR
IsoMedia import - track ID 1 - Audio (SR 48000 - 2 channels)
Saving /Users/lhawkins/Desktop/FIGHTCLB/20real-2pass-MultiHex.mp4: 0.500 secs Interleaving
Encoding completed on Wed Sep 14 13:40:05 PDT 2005

[lhawkins]

BTW, I ripped the DVD with MTR in full disc mode and am feeding FFMPEGX the VIDEO_TS folder as a source.

[major]

Unfortunately, I need the actual file to test. So far the DVD I have had no sync problems.

[magicite]

Unfortunately, I need the actual file to test. So far the DVD I have had no sync problems.

Would it be possible for you to list what DVDs you have available to test? That way, if one of us who are having problems has an identical DVD, we can test things.

Also, same problem when encoding 'Aladdin.'

Here's some more information that may be helpful to you:
Dual 2.0 GHz G5 w/ 1.5 GB RAM running 10.4.2
Converting from non-boot (internal SATA) drive (ie, the DVD folder that MTR ripped) to an h264 file (going onto the same hard drive).
Video Tab:
720x304 (after crop)
NTSC FILM framerate
Audio Tab:
96 kbit / sec VBR AAC stereo, 48,000 Hz
Options Tab:
Nothing checked
qmin=qmax=22
scaling method: bicubic spline
ME function: exhaustive

Here's my list of currently-ripped DVDs (on my HD, ready to be converted):
(list)

[major]

Please send the info to major4@mac.com

[Metallirat]

We need the same file. Some sort of public domain movie. Something from Archive.org might work. Im going to download a mpeg2 movie from there and try it out

[Mechcozmo]

I will convert a movie and send links to you major. MPEG2->H.264 along with the output. Lemme get it going so you can start the download... I have 50KB/sec up here, so it may take you a bit.

By my guess you will be able to start your 'work' tomorrow night. Tonight to convert, tomorrow to download... (177MB projection-- I turned bitrate down to 700 and hope that doesn't mess anything up) So, all in all major, we shall hopefully get an answer to this mess soon. For the record, I'm using 0.0.9t r3.

[major]

Thank you.

[Mechcozmo]

Public announcement: Video encoding went bad, going to restart. Bleach.

I will email you the link to the server and the output file whenever it finishes... :-\

[magicite]

Major and I are also collaborating on this problem, doing more or less the same thing. Hopefully, we can get this issue cleared up soon!

[Metallirat]

Major and I are also collaborating on this problem, doing more or less the same thing. Hopefully, we can get this issue cleared up soon!

Ditto. I'm just going to be patient and wait at this point. If anyone wants to try encoding the same thing go to http://www.archive.org/details/vice_presidential_debate_10_5_04

get the mpeg2 version. For me it was in sync at first but right near the end it started getting out of sync.

[major]

Thank you. I'm finding time this weekend to do all tests.

[Mechcozmo]

Thank you. I'm finding time this weekend to do all tests.

I saw in the logs you downloaded the file... I assume the file is suitable for testing, etc? Just making sure.

[zachk]

I'm not very knowledgeable in these matters, but it seems to me that a very likely reason for the syncing problems is that for h.264 the video and audio aren't encoded at the same time. If you have 2 pass selected, mencoder will encode just the video twice, and then ffmpeg will take over and encode just the audio.

Major, why doesn't mencoder do both the audio and video at the same time? And, is this the source of the syncing issues?

[major]

Yes, the syncing problem is related to the fact that mencoder does not mux. This is because mencoder is unable to mux mp4. However, this is not a valid enough reason for a sync problem to appear, so there is hope to solve it.

[BAL_050905]

http://www.archive.org/details/vice_presidential_debate_10_5_04

get the mpeg2 version. For me it was in sync at first but right near the end it started getting out of sync.

I'm not Engrish speaker. thus, could not find LipSync prob.
Can you tell me "near the end" in Time? like 00:30:44? So I can find it more carefully. Anyway, I agree "We need the same file".

A/V Sync Test for ffmpegX0.0.9t R3
0.Source File vice_presidential_debate_10_5_04_3.mpeg
1.Results
-Source; 1.49GB ; 33min59.85sec(2039.85);
-009tr3; 191.4MB ; 33min59.84sec(2039.84); -0.01sec(99.99951%)
2.009tr3Settings
-V_tab;1078kbps,640x480,NTSC
-A_tab;96kbps,48000khz,
-F_tab;Deinterlace,Denoise,x264deblock
-O_tab;CABAC,b-frames,Constant bitrate,fast bilinear,Diamond
3.Source File info from MPEGstreamclip
-Duration: 0:33:59
-Video Tracks:
--224 MPEG-2, 720 × 480, 4:3, 29.97 fps, 6.00 Mbps, upper field first
-Audio Tracks:
--192 MP2 stereo, 48 kH, 192 kbps
4.Enviroment
-Machine; PowerMac G5 2Ghz dual(2004),mem=1GB
-OS; Mac OSX 10.4.2,QuickTime Player 7.0.2 Pro
-Encoder; ffmpegX0.0.9t-R3

###issue2###
Does anyone know typical AviSynth settings? I think I can find similar options for -vop chain.
This Windows softwares seems doing almost same as ffmpegX's MEncoder.
MeGUI; http://forum.doom9.org/showthread.php?t=96032
this uses
"mencoder.exe" "D:\DVDs\DVDVolume\VIDEO_TS\re-trailer.avs" -ovc x264 -x264encopts bitrate=700:subq=4:i4x4 -o "D:\DVDs\DVDVolume\VIDEO_TS\test.264" -of rawvideo
no "-ni",no "-vop".
So AviSynth (.avs) may have hints.
YAMB; http://forum.doom9.org/showthread.php?s=&threadid=93927
This one seems doing quite same as ffmpegX's mp4box.

[magicite]

Hey Major,
Any progress so far? I know you said you were going to work on it this past weekend. Keep up the good work!

[major]

I'm encoding with many different settings and waiting to compare results.

[DonSqueak]

Hi major,

not wanting to bother you in any way, only still having the same problem: is this whole out-of-sync thing in any way related to my VOB 5.1 to AVI 5.1 out-of-sync problem?

https://www.videohelp.com/forum/viewtopic.php?t=277746

Cheers, Martin.

[major]

I tried encoding vice_presidential_debate_10_5_04_3.mpeg to NTSC, no frames were duplicated or skipped, and the MP4 file seem to play in sync here (using mplayer to playback). Source was 00:30:50.5, Target was 00:30:50.5. I'm currently trying with some full DVD.

[TheBum]

To add another possibility, Star Wars Episode IV (Region 1 NTSC Widescreen) encoded in H264/AAC exhibits about one second of video lag by the time it reaches the end.

[Mechcozmo]

major, what is your computer situation like? If someone here has OS X server than we could set up an Xgrid array to help you out. I have a 24/7 server that would be more than happy to help you out.

[major]

I found a DVD with the infamous one-second lag. I'm now researching about possible solutions.

[bigbubba]

Been waiting for 2 days after registering so I could post this.

To add to the confusion, I haven't had any success with encoding H.264. The only videos I've tried so far (National Treasure and a sitcom episode, both from DVDs) have shown sync problems for me when playing in both QT and VLC.

The National Treasure H.264 video (2+ hour movie) was 45 seconds shorter than the audio track (AAC 128kbit). The 25 minute sitcom was about a second behind the audio. When doing the movie, I got the "I duplicate frame(s)!" in the beginning (one warning, not continuous) but that's it. Output was unconstrained and frame rate the source rate (29.97).



Is it something with the way the VOB files are read?

For example, for the sitcom, demuxing via "Demux to M2V and AC3" in MPEG Streamclip 1.3.1 with Apple's MPEG-2 playback component installed gives me a 0:23:32:29 length video clip and a 0:23:33:00 length audio clip instead of the VOB's 0:23:22:00 length for both (VOB and M2V lengths reported by MPEG Streamclip, audio length reported by FCE HD). The AC3 audio was converted to AIFF by bd4go, and it was still the same length.

Another interesting observation... importing both the M2V AIFF files produced by MPEG Streamclip ("Demux to M2V and AC3", then AC3 converted to AIFF by bd4go) into FCE HD gives 0:23:30:06 for the m2v track and 0:23:33:00 for the audio track, for a more pronounced length difference.

When importing the original VOB file into FCE HD, the reported length is 23:31:18. I re-demuxed the file again, using different settings, and then imported the finished files into FCE HD for the length report. For comparison, the original length of the VOB file (as reported by MPEG Streamclip) is 23:32:28.

"Demux to M2V and AC3"
-M2V was 23:30:06
-AC3 to AIFF was 23:33:00

"Demux to M2V and AIFF"
-M2V was 23:33:01
-AIFF was 23:33:00

"Demux to Unscaled M2V and AIFF"
-results were the same as above

"Demux to Headed M2V and AIFF"
-results were the same as above
-I should have checked all the way through, but I put the M2V file from this demux back into MPEG Streamclip and it reported the length as being 23:32:29 vs. FCE HD's 23:33:01.



Demuxing the file with ffmpegX gives me similar results: a 23:30:06 M2V track and a 23:32:29 AIFF track (converted from AC3) as reported in FCE HD.

I was curious if any of this inconsistency was caused in part by Apple's MPEG-2 component, so I uninstalled it, restarted, and demuxed the file again in FFmpegX. After reimporting the files into FCE HD, I got the same length (23:30:06/23:32:29).

I think the problem lies with the demuxing, or the way the program reads the file length. But then that would mean non-H.264 rips would be out of sync as well, and it sounds like they are fine.

[major]

The problem is in the video stream, and is caused by mencoder. I'm researching for solutions.

[bigbubba]

Anything I can do to help out? Is it a known issue with mencoder, or is it specific to the Mac version? (it looked like mencoder and mplayer is also available for the PC.)

[major]

It should be the same on PC. I'm checking about possible solutions.

[JonStew1983]

major,

Any news on this, like bigbubba I signed up just to see if there was any news. I've been googling for weeks and if I ever get the time I will probably post something on the mencoder list with a link to sample files but I've been really busy lately and don't have very good net access. So far every DVD I've tried to encode has been behind some fraction of a second on the sound, I can't be more accurate than that though because I have trouble telling whether it's ahead or behind most of the time much less guessing how much. I can say it's behind for sure in these cases though. If there is anything I can test or if you want to see the command lines I'm using just let me know. I just don't know what else to try at this point.

Thanks,
Jonathan

[major]

Apparently this happen only with NTSC FILM sources. When audio is not encoded at the same time as the video, the demuxer framerate detection seem to fail during the first one or two seconds.