[Elphel-support] mono video output postprocessing

[support-list]

Gergely,

with the full frame the maximal fps in stereo configuration can be up to 7.2 - provided image quality/number of fine details on the image allow to fit the bandwidth of sesnoding the images over the network (up to 10MB/s) or record it to HDD/SSD (up to 16MB/s). You can reach 10fps if you slightly reduce the FOV - vertically, horizontally, or both.
You can get higher  FPS if you use to separate cameras synchronized - that is possible with 10369 boards.
You will be able to get to even higher FPS when our new NC393 camera will be ready, but it will be ready only by the mid-2014.

Andrey

---- On Mon, 05 Aug 2013 01:43:54 -0700 Gergely Debreczeni<gergely.debreczeni at kishonti.net> wrote ---- 

 > Dear Andrey ! 
 >  
 >  Thanks for these pointers. We will try your advices, however the problem with 
 > JP4 is that it is too big. We would need a minimum of 10 fps @ half resolution, 
 > which we found is not feasible for this camera. 
 >  
 > Anyway, will so how far we can get. 
 > Cheers, 
 > Gergely 
 >   
 > ________________________________________ 
 > De : support-list <support-list at support.elphel.com> 
 > Envoyé : jeudi 1 août 2013 06:14 
 > À : Gergely Debreczeni; Elphel List 
 > Objet : Re: [Elphel-support] mono video output postprocessing 
 >  
 > ---- On Wed, 31 Jul 2013 02:58:16 -0700 Gergely Debreczeni<gergely.debreczeni at kishonti.net> wrote ---- 
 >  
 >  >   Hello, 
 >  > 
 >  >  Now we use our stereo camera in mono mode, so the resulted 
 >  >  8bit grayscale video's pixels corresponds to the color value of the CCD 
 >  >  in 
 >  > 
 >  >   G G 
 >  >   R B 
 >  > 
 >  
 > Gergely, it is actually 
 > GR 
 > BG 
 > greens are diagonal. 
 >  
 >  >  layout - if I'm right. 
 >  >  I'm just wondering: Is there any easy way (preferably in Linux) 
 >  >  to convert this format into a video of half size, half width where one pixel 
 >  >  corresponds to some function (intensity/average/etc..) of the above four pixel ? 
 >  
 > It will not give you much - the center of R and B colors is shifted, so to do that correctly you need de-Bayer or demosaic of your data. And I strongly recommend you using JP4, not JPEG if you want raw Bayer from the camera. 
 > Additionally all the lenses have aberrations, including chromatic, and blue color resolution is significantly worse than for green (or even red) - it is the same as for the human eyes. So if you want to easy way to process data for 3d reconstruction - forget about blue and red - just use (and calibrate distortions) green only. 
 >  
 >  > 
 >  >  I was looking for some filter in ffmpeg, but found none which could perform 
 >  >  this task. 
 >  
 > Yes, it is so because there is no easy and useful way to do that. 
 >  
 > Andrey 
 >  
 >  
 > _______________________________________________ 
 > Support-list mailing list 
 > Support-list at support.elphel.com 
 > http://support.elphel.com/mailman/listinfo/support-list_support.elphel.com 
 >