Hi, Has anyone tried using mencoder with the -edl option to edit out parts of the input stream? I'm finding that after a few edits, there's a loss of a/v sync. Is this a known problem? Thanks, - Mart
Maurice Cooper wrote:
Hi, Has anyone tried using mencoder with the -edl option to edit out parts of the input stream? I'm finding that after a few edits, there's a loss of a/v sync. Is this a known problem?
I've been using a version of the EDL code from a month or two ago and found it works very well. The problem I was having with A/V sync was that the edits seemed to be screwing up a filter, but actually I wasn't using the filter correctly in the first place. An edit will often break an audio packet and the broken packet will probably print an error. Since my edits have been during quiet sections I haven't heard it cause any noise or any noticeable effects. There is a problem when you attempt to edit out a section in which the pts's have wrapped around back to zero. The skipping code will get confused because the end of the skipped section will have a pts lower than the beginning. The one time this happened to me, I had to make a new version of the unedited original so its pts would start at zero and redo the edits. This won't be a common situation though.
On 4/25/05, Scott W. Larson <scowl@pacifier.com> wrote:
I've been using a version of the EDL code from a month or two ago and found it works very well. The problem I was having with A/V sync was that the edits seemed to be screwing up a filter, but actually I wasn't using the filter correctly in the first place.
An edit will often break an audio packet and the broken packet will probably print an error. Since my edits have been during quiet sections I haven't heard it cause any noise or any noticeable effects.
There is a problem when you attempt to edit out a section in which the pts's have wrapped around back to zero. The skipping code will get confused because the end of the skipped section will have a pts lower than the beginning. The one time this happened to me, I had to make a new version of the unedited original so its pts would start at zero and redo the edits. This won't be a common situation though.
I'm editing out three sections from 30 minute MPEG-4 files encoded using mencoder - removing ad breaks. By the end of the video, just before the last edit, the audio is approximately half a second out. Could it be because the files are encoded using vqscale=1? - Mart
On Mon, Apr 25, 2005 at 12:38:12AM +0100, Maurice Cooper wrote:
On 4/25/05, Scott W. Larson <scowl@pacifier.com> wrote:
I've been using a version of the EDL code from a month or two ago and found it works very well. The problem I was having with A/V sync was that the edits seemed to be screwing up a filter, but actually I wasn't using the filter correctly in the first place.
An edit will often break an audio packet and the broken packet will probably print an error. Since my edits have been during quiet sections I haven't heard it cause any noise or any noticeable effects.
There is a problem when you attempt to edit out a section in which the pts's have wrapped around back to zero. The skipping code will get confused because the end of the skipped section will have a pts lower than the beginning. The one time this happened to me, I had to make a new version of the unedited original so its pts would start at zero and redo the edits. This won't be a common situation though.
I'm editing out three sections from 30 minute MPEG-4 files encoded using mencoder - removing ad breaks. By the end of the video, just before the last edit, the audio is approximately half a second out.
Could it be because the files are encoded using vqscale=1?
I'm totally confused what vqscale could have to do with a/v sync.... Anyway, are you using -hr-edl-seek or not? Knowing this might help. Rich
On 4/25/05, Rich Felker <dalias@aerifal.cx> wrote:
On Mon, Apr 25, 2005 at 12:38:12AM +0100, Maurice Cooper wrote:
Could it be because the files are encoded using vqscale=1?
I'm totally confused what vqscale could have to do with a/v sync....
Anyway, are you using -hr-edl-seek or not? Knowing this might help.
I've tried this, and it skips to the start of the video, and then seg faults. - Mart
Sorry for the late reply. On Mon, Apr 25, 2005 at 11:42:15AM +0100, Maurice Cooper wrote:
On 4/25/05, Rich Felker <dalias@aerifal.cx> wrote:
On Mon, Apr 25, 2005 at 12:38:12AM +0100, Maurice Cooper wrote:
Could it be because the files are encoded using vqscale=1?
I'm totally confused what vqscale could have to do with a/v sync....
Anyway, are you using -hr-edl-seek or not? Knowing this might help.
I've tried this, and it skips to the start of the video, and then seg faults.
That sounds bad, could you give more information about the segfault? Params, what video, what edl, maybe even a backtrace. - ods15
On 5/6/05, Oded Shimon <ods15@ods15.dyndns.org> wrote:
Sorry for the late reply.
On Mon, Apr 25, 2005 at 11:42:15AM +0100, Maurice Cooper wrote:
On 4/25/05, Rich Felker <dalias@aerifal.cx> wrote:
On Mon, Apr 25, 2005 at 12:38:12AM +0100, Maurice Cooper wrote:
Could it be because the files are encoded using vqscale=1?
I'm totally confused what vqscale could have to do with a/v sync....
Anyway, are you using -hr-edl-seek or not? Knowing this might help.
I've tried this, and it skips to the start of the video, and then seg faults.
That sounds bad, could you give more information about the segfault?
Params, what video, what edl, maybe even a backtrace.
I think so, but it'll take a little time to recreate what I was doing. How do I get a backtrace? - Mart
On Mon, 9 May 2005 23:12:12 +0100 Mart <pyntur@gmail.com> wrote:
How do I get a backtrace?
Completely explained: http://mplayerhq.hu/DOCS/HTML/en/bugreports.html
participants (6)
-
Mart -
Maurice Cooper
-
Oded Shimon -
RC -
Rich Felker -
Scott W. Larson