Airw0lf
15th January 2002, 08:11
Hi, I'm wondering if anyone could help me sort out a weird problem I'm having. First, here's my setup
I capture video from TV using virtualdub (480 x 576, Pinnacle MJPEG codec , PCM CD Quality audio.) Now when I capture, i usually end up with several files of about 10 mins length (e.g., when an ad break comes on, I stop recording and start once the show's back.)
In order to make life easy I load up the files (appending .avis) in virtualdub and do all my editing there before frameserving to CCE 2.50. (Also apply a few filters like deinterlace etc) That way, when the encoding is finished, I have the final full, edited video.
I DON'T encode audio in CCE. I use TMPGEnc as a TooLAME gui to convert to .mp2 separately, and I mux with bbMPEG.
My problem is, I usually notice weird sync problems beyond the first 20 mins of my SVCDs. But the sound actually never really gets significantly ahead/behind the video, but it's really obvious when people are speaking that their lips don't match what you hear. It's ALMOST tolerable, but not quite right.
From my tests, it seems if i encode my 10 min segments separately, everything is synced perfectly. So what gives? Is it something to do with the way I'm concatenating in Virtualdub? (I notice with some captures, virtualdub automatically offsets the sound by X ms in order to keep sync, even though 0 frames are dropped. Should I account for this in cutting/editing? The avis themselves always playback beautifully synced) Is it the muxing? btw I NEVER get any PTS/DTS underflows.
Is anyone doing something similar to me? I'd really appreciate some help.
I capture video from TV using virtualdub (480 x 576, Pinnacle MJPEG codec , PCM CD Quality audio.) Now when I capture, i usually end up with several files of about 10 mins length (e.g., when an ad break comes on, I stop recording and start once the show's back.)
In order to make life easy I load up the files (appending .avis) in virtualdub and do all my editing there before frameserving to CCE 2.50. (Also apply a few filters like deinterlace etc) That way, when the encoding is finished, I have the final full, edited video.
I DON'T encode audio in CCE. I use TMPGEnc as a TooLAME gui to convert to .mp2 separately, and I mux with bbMPEG.
My problem is, I usually notice weird sync problems beyond the first 20 mins of my SVCDs. But the sound actually never really gets significantly ahead/behind the video, but it's really obvious when people are speaking that their lips don't match what you hear. It's ALMOST tolerable, but not quite right.
From my tests, it seems if i encode my 10 min segments separately, everything is synced perfectly. So what gives? Is it something to do with the way I'm concatenating in Virtualdub? (I notice with some captures, virtualdub automatically offsets the sound by X ms in order to keep sync, even though 0 frames are dropped. Should I account for this in cutting/editing? The avis themselves always playback beautifully synced) Is it the muxing? btw I NEVER get any PTS/DTS underflows.
Is anyone doing something similar to me? I'd really appreciate some help.