(semi-ot) Camcorder frequency drift on multi-cam captures

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]



Hello,
I need to write an application that captures two video streams simultaneously.
I'm using two consumer camcorders and bttv cards on v4l.

I'm using a two thread model (each thread captures one stream) and I time 
stamp each frame.
My goad is to track a few points in the video streams which are then used
 for 3d trajectory reconstruction.
Since I cannot use genlocks I have to resort to other "tricks" to minimize
 synchronization errors. (frequency drifts between the two cameras (I'm using 
PAL so the nominal freq. is 50Hz))
Since I'm only analyzing trajectories, I thought the easies thing to do would
be to timestamp the frames and interpolate coordinates linearly (based on
current x,y previous frame's x,y and time stamp).

What I was wondering if someone of you has an idea what kind of drifts
 consumer cameras experience in the real world (eg +/-1% of 50Hz ?) and
if these drifts are pretty constant or vary due to temperature changes etc.
Anyway common sense would tell me that the drift would be pretty small
and constant so I guess linear interpolation is more than adequate.

thanks for infos,
cheers,
Benno

http://www.linuxaudiodev.org The Home Of Linux Audio Development





[Index of Archives]     [Linux DVB]     [Video Disk Recorder]     [Asterisk]     [Photo]     [DCCP]     [Netdev]     [Xorg]     [Util Linux NG]     [Xfree86]     [Free Photo Albums]     [Fedora Users]     [Fedora Women]     [ALSA Users]     [ALSA Devel]     [Linux USB]

Powered by Linux