Hi All, We have been running our systems with the increased buffer size as suggested by Tom, and this gave a good improvement. For another performance boost I suggest adjusting the nice level and scheduling priority of the process which receives the camera data. You might need to adjust /etc/security/limits.conf to allow non-root owned processes to adjust their nice-ness and scheduler priority. John On Thu, Apr 17, 2014 at 3:51 PM, Patrick Doyle <wpdster@xxxxxxxxx> wrote: > After all of that, it turns out that there was a bug in the firmware > in my camera (which has firmware that is newer than that available on > the Point Grey website). Point Grey sent me a new firmware image > (1.31.3.0) and the timestamps now make sense. > > Now, on to debugging the lost packets problem… I anticipate that Tom's > suggestion that I increase the network buffer sizes will go a long way > towards making that problem go away. > > Emmanuel, > Would you object to a (yet to be written, let alone tested and/or > debugged) patch to add a packet-resend flag to the gst plugin? Now > that I have timestamps working, I am less concerned about dropped > frames, as I will be able to detect those events from the camera > timestamps. > > --wpd > > On Wed, Apr 16, 2014 at 9:39 AM, Patrick Doyle <wpdster@xxxxxxxxx> wrote: > > Actually, after writing my long email, my long email, I (finally) > > decided to look at the PointGrey documentation. Scanning through > > their Technical Reference manual, I see that they offer the ability to > > encode the timestamp, gain, exposure, brightness, etc… in the first N > > bytes of each image transmitted. They then go on to document the > > timestamp format that gets encoded in the image as being a 32 bit > > integer, where the top 7 bits represent seconds, and wrap around every > > 128 seconds (presumably incrementing the high order 32 bit register > > when they do so, as I observed). The next 13 bits contain a 13-bit > > "Cycle Count" in units of 125 uS that increments from 0 to 7999. And > > the lower order 12 bits contain a "Cycle Offset", which I don't > > understand yet. It seems reasonable that the out-of-band timestamp > > would follow the same format as the in-band data. > > > > Scanning briefly the 3 data logs I collected, I see that the 13 bit > > Cycle Count value increments by 100 for each frame received when I > > (think I) configured the camera at 60 fps, and by 200 for my 30fps > > log. > > > > Hmmm… > > > > --wpd > > > > > > On Wed, Apr 16, 2014 at 9:22 AM, Emmanuel Pacaud <emmanuel@xxxxxxxxx> > wrote: > >> Hi, > >> > >> Le mercredi 16 avril 2014 à 09:02 -0400, Patrick Doyle a écrit : > >>> I can think of 3 possibilities: > >>> 1) Handle timestamps from BlackFly cameras as a special case within > >>> the Aravis framework. > >>> 2) Extend the Aravis framework to allow it to report raw timestamps > >>> received from a camera and write my own gstreamer plugin to convert > >>> Blackfin timestamps to ns. > >>> 3) Do something smarter suggested by one of you. > >> > >> A last possibility is to contact PointGrey support and submit what you > >> have observed. It can be a bug in Aravis, but also in the camera > >> firmware. If it is the latest, a fix in the firmware would be the best > >> solution. > >> > >> Cheers, > >> > >> Emmanuel. > >> > >> > >