Selur's Little Message Board

Full Version: Audio/video out of sync in output
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3
Hello,
I am trying to encode and deinterlace some PAL AVIs to x264 and I'm noticing that for the current file, audio and video drift out of sync towards the middle of the converted file. IM'd to Selur is a 10s slice (~minute 34) of the original AVI file (1h 25m) in which I am noticing the sync issue. Please also find the logs attached.

Thanks
Will look at it today after work or tomorrow and report back.

Cu Selur
1st thing: Why do you capture interlaced content to a format like Lagarith which does not support interlaced content?
2nd thing: Can you reproduce the issue with the sample you shared with me?
I don't have any sync issues reencoding the file with the settings the debug output has.
Will send you a link to the files I encoded,..

Cu Selur
(31.07.2020, 19:21)Selur Wrote: [ -> ]1st thing: Why do you capture interlaced content to a format like Lagarith which does not support interlaced content?
2nd thing: Can you reproduce the issue with the sample you shared with me?
I don't have any sync issues reencoding the file with the settings the debug output has.
Will send you a link to the files I encoded,..

Cu Selur

Thanks,
1. Unfortunately, no one told me that. I had a choice between Huffyuv and Lagarith and had some issues with Huffyuv.
2. I don't have sync issues with the 10s slice either. It's in the context of the whole file that the issue occurs.

Anything else you want me to try? Also, does capturing in Lagairth result in a degradation in capture quality if the source is interlaced?
Quote:Also, does capturing in Lagairth result in a degradation in capture quality if the source is interlaced?
Problem with Lagarith from my experience is:
a. any time I used it I can into problems
b. it does not support interlaced content at least it does not properly signal that the content is interlaced
No clue whether it breaks anything when used for interlaced content, I hope not, but you need to overwrite the scantype when handling it.
-> better use FFV1 or UT Video Codec Suite both should support interlaced content iirc.

That said, problem is: I need a sample which allows me to reproduce the issue otherwise I can't tell you where the problem is. Smile

Cu Selur
(01.08.2020, 08:44)Selur Wrote: [ -> ]
Quote:Also, does capturing in Lagairth result in a degradation in capture quality if the source is interlaced?
Problem with Lagarith from my experience is:
a. any time I used it I can into problems
b. it does not support interlaced content at least it does not properly signal that the content is interlaced
No clue whether it breaks anything when used for interlaced content, I hope not, but you need to overwrite the scantype when handling it.
-> better use FFV1 or UT Video Codec Suite both should support interlaced content iirc.

That said, problem is: I need a sample which allows me to reproduce the issue otherwise I can't tell you where the problem is. Smile

Cu Selur
First of all, I really appreciate your quick response every time.
Second, I understand you need the actual sample, but it's ~40GB. I've been setting the scan type to TFF manually every time. Is there anything I need to change in Hybrid previously used for NTSC to encode PAL?
Quote:Is there anything I need to change in Hybrid previously used for NTSC to encode PAL?
NTSC -> PAL is mainly:
a. adjusting the resolution
b. adjusting the frame rate
the only tricky part is the decision on how to adjust the frame rate.
Personally I would either:
a. just change the frame rate and reencode the audio (stretching the audio without reencoding is also possible when using a mkv container)
b. interpolate the missing frames to compensate the higher frame rate and leave the audio as it is.

Another idea: The problem might be with handling the avi container. -> does it help to remux the container to mkv using MKVToolnix before reencoding it?

Cu Selur
(01.08.2020, 09:27)Selur Wrote: [ -> ]
Quote:Is there anything I need to change in Hybrid previously used for NTSC to encode PAL?
NTSC -> PAL is mainly:
a. adjusting the resolution
b. adjusting the frame rate
the only tricky part is the decision on how to adjust the frame rate.
Personally I would either:
a. just change the frame rate and reencode the audio (stretching the audio without reencoding is also possible when using a mkv container)
b. interpolate the missing frames to compensate the higher frame rate and leave the audio as it is.

Another idea: The problem might be with handling the avi container. -> does it help to remux the container to mkv using MKVToolnix before reencoding it?

Cu Selur
Sorry, I meant just for encoding to x264. Framerate and resolution were adjusted for capture using virtualdub.
The other 4 PAL AVIs (45m files) I have encoded so far seem to be fine. Does it help if I reencode the faulty one (1h 25m file) and provide the logs?
Quote:Framerate and resolution were adjusted for capture using virtualdub.
I hope your input is already interlaced and not telecined, otherwise capturing interlaced and using QTGMC (or any other deinterlacer) is probably wrong, since it would retain the 3:2 pattern in the progressive frames. Smile

Quote:Does it help if I reencode the faulty one (1h 25m file) and provide the logs
Isn't that what you did to create the debug output you attached to the first post?

Cu Selur
(01.08.2020, 10:34)Selur Wrote: [ -> ]
Quote:Framerate and resolution were adjusted for capture using virtualdub.
I hope your input is already interlaced and not telecined, otherwise capturing interlaced and using QTGMC (or any other deinterlacer) is probably wrong, since it would retain the 3:2 pattern in the progressive frames. Smile

Quote:Does it help if I reencode the faulty one (1h 25m file) and provide the logs
Isn't that what you did to create the debug output you attached to the first post?

Cu Selur
Oh man!!! The footage was captured using a Tevion capture card and a panasonic camcorder. Isn't the 10s sample enough for identifying? I can see the combing effect though which I've read is a good indicator of interlaced content.
No, that's the log for the 10s sample. I will provide the new logs in the morning.
Pages: 1 2 3