Yes, that's my first guess, and I try to fix it with "-async num" or
"-vsync 1", I get the idea from:
http://lzone.de/cheat-sheet/ffmpeg
while for my case, it didn't related with different clock crystal
(actually the two stream comes from same capture device), it is because
I incorrectly use very different time stamp:
Input #0, alsa, from 'hw:3':
Duration: N/A, start:*1479202954**.916662*
Input #1, video4linux2,v4l2, from '/dev/video2':
Duration: N/A, start:*31.451263*
I guess it related with the accuracy of float calculation during time
stamp scaling/resample, maybe Carl could give the reason?
Thanks,
Luo Xin
On 2016年11月18日 12:37, Reuben Martin wrote:
On Wednesday, November 16, 2016 7:59:44 PM CST Luo Xin wrote:
- capture video from v4l2 device
- capture audio from alsa device
The drift is going to happen to a certain extent no matter what. You are
sourceing from two separate pieces of hardware that do not share a common
hardware clock crystal.
-Reuben
_______________________________________________
ffmpeg-user mailing list
[email protected]
http://ffmpeg.org/mailman/listinfo/ffmpeg-user
To unsubscribe, visit link above, or email
[email protected] with subject "unsubscribe".
_______________________________________________
ffmpeg-user mailing list
[email protected]
http://ffmpeg.org/mailman/listinfo/ffmpeg-user
To unsubscribe, visit link above, or email
[email protected] with subject "unsubscribe".
