In your test with 2 or 3 inputs, obtain fluid video in each camera? or they is fluid in a moment and stop in lags in other moment? Fabio > -----Mensaje original----- > De: Ed Okerson [mailto:eokerson@xxxxxxxxxxxxxxxx] > Enviado el: domingo 20 de julio de 2003 21:12 > Para: video4linux-list@xxxxxxxxxx > Asunto: RE: multiple open()'s on the same device > > > On Sun, 20 Jul 2003, Ryan Ayers wrote: > > > > From: Christian Hack > > > I still stand by my statement. Does the windows version > capture full motion > > > on all 4 channels. The only simultaneous capture cards I > have seen have a > > > different capture chip or have more than one capture chip fitted. > > > > > > Feel free to prove me wrong though. ;-) > > > > > > CH > > > > I've spoken with Gerd about this previously. The only way > todo this with > > the current linux bttv driver, is to switch inputs (very slowly). > > Generally, this means, grab 1 frame, switch and skip 5 > frames (waiting for > > the video device to settle down). This gives a realistic > throughput of > > about 2 fps switching between 2 inputs. This is totally > unacceptable for a > > surveillance solution. There is some trickery going on in > the windows > > driver, because the exact same card with the windows driver > can capture > > around 10-15fps across 2 inputs. > > I have successfully used the package at > http://motion.sourceforge.net to > get much better switching between ports on bttv based cards. > It uses the > vloopback device to attach different inputs to virtual > /dev/video devices. > I had no problems getting 13-15fps from two inputs, and > 8-10fps from three > inputs. > > Ed Okerson > > >
