dust12
25th November 2002, 08:38
Hi all,
Whenever I capture long (about 1 hour) captures using my DC10+ card under Windows XP Pro with Studio 8.3 and subsequently encode the MJPEG-files using CCE, audio and video get out of sync. The audio part is about half a second per hour playtime faster than video. I tried it with both CCE 2.50 and 2.66.06 (trial). It doesn't matter either whether audio is encoded within CCE or seperately using toolame. The MJPEG files (before encoding) are perfectly synchronous.
If I encode using TMPGEnc (version 2.58) instead, the result is perfectly synchronous.
Also, if I capture under W98SE using VirtualDub or AVI_IO (both produce sequential AVI1 files, whereas Studio under XP Pro produces OpenDML 1.02 files) and feed CCE with an AviSynth script that sticks the 2GB-segments together, the result is again in sync using both CCE and TMPGEnc.
It is important to know that a certain A/V drift is inavoidable using the DC10+ because audio is recorded by the soundcard. The capture program can correct this drift by correcting the videostream. AVI_IO and VirtualDub drop frames if the video part is too fast or insert empty frames (that are flagged as "dropped frames") if it is too slow. So, using AVI_IO or VirtualDub, I can see about 12 dropped frames an hour - these are artifially added frames to slow down video so it keeps in sync with audio. 12 frames also last about half a second, which is the drift I observed above.
A possible reason for this problem is that maybe OpenDML 1.02 offers more advanced features that get along without actually adding empty frames (like tagging the video frames directly with timecodes), and maybe CCE ignores these features? This would mean CCE had a problem with its OpenDML implementation, whereas the really inexpensive TMPGEnc gets along with it well? This seems pretty surprising to me, considering the high price of CCE.
Did anybody observe something similar, or maybe this is an already known problem of CCE?
Thanks in advance, Andy
Whenever I capture long (about 1 hour) captures using my DC10+ card under Windows XP Pro with Studio 8.3 and subsequently encode the MJPEG-files using CCE, audio and video get out of sync. The audio part is about half a second per hour playtime faster than video. I tried it with both CCE 2.50 and 2.66.06 (trial). It doesn't matter either whether audio is encoded within CCE or seperately using toolame. The MJPEG files (before encoding) are perfectly synchronous.
If I encode using TMPGEnc (version 2.58) instead, the result is perfectly synchronous.
Also, if I capture under W98SE using VirtualDub or AVI_IO (both produce sequential AVI1 files, whereas Studio under XP Pro produces OpenDML 1.02 files) and feed CCE with an AviSynth script that sticks the 2GB-segments together, the result is again in sync using both CCE and TMPGEnc.
It is important to know that a certain A/V drift is inavoidable using the DC10+ because audio is recorded by the soundcard. The capture program can correct this drift by correcting the videostream. AVI_IO and VirtualDub drop frames if the video part is too fast or insert empty frames (that are flagged as "dropped frames") if it is too slow. So, using AVI_IO or VirtualDub, I can see about 12 dropped frames an hour - these are artifially added frames to slow down video so it keeps in sync with audio. 12 frames also last about half a second, which is the drift I observed above.
A possible reason for this problem is that maybe OpenDML 1.02 offers more advanced features that get along without actually adding empty frames (like tagging the video frames directly with timecodes), and maybe CCE ignores these features? This would mean CCE had a problem with its OpenDML implementation, whereas the really inexpensive TMPGEnc gets along with it well? This seems pretty surprising to me, considering the high price of CCE.
Did anybody observe something similar, or maybe this is an already known problem of CCE?
Thanks in advance, Andy