EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  <20072008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024  Index 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  <20072008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: Re: Frame rate performance
From: "D. Peter Siddons" <[email protected]>
To: Emmanuel Mayssat <[email protected]>
Cc: [email protected]
Date: Thu, 11 Oct 2007 04:17:56 -0400
Hi Emmanuel,
I'm always nervous of throwing away information! You're probably right in that it doesn't matter for your applications, but in a more quantitative situation (e.g. computed tomography), where you might want the frame rate, I would avoid compression if it was not lossless. And as Mark says, so what if I need another Ethernet card!
Pete.


Emmanuel Mayssat wrote:
On Wed, 2007-10-10 at 19:22 -0400, D. Peter Siddons wrote:
Doesn't the Axis server do some (lossy?) compression?
Pete.


Yes...
I was actually dumbfounded when I read the bandwidth requirement for the
Gig-e camera. It is actually in the name! You need a gigabyte network. Those camera don't use compression at all, hence the high bandwidth
requirement.


With the axis, you can use motion JPEG and MPEG-4.
Motion JPEG is a succession of JPEG images (lossy compression
algorithm). MPEG-4 not only compress the images individually, but also
between successive images (stream of images). So with MPEG-4 you have
even more losses.

Now, the question becomes should I care about the loss of information in
the video stream. My answer: it depends on the amount of loss. JPEG
(motion jpeg) losses are such that they cannot be detected by the human
eye. JPEG's fundamental idea is to reduce things that appear invisible
to the human eye and by doing so, tremendously increase the compression
ratio. Those losses would be detected by a computer though, a computer
can carefully look at the RGB value of each individual pixel.

Now, should I care about loss? Probably not for electron beam centering
or surveillance.


For beam centering:
1/ my image is mostly black (image is mostly uniform, JPEG compression
ratio is huge.)
2/ what I care about is not the image itself but statistics (loss/gain
of intensity on a few pixel doesn't matter much and is averaged out) 3/ image is mostly static (e- beam stay at the same place), so I could
use mpeg4 for even more compression
4/ frame rate is configurable (do you really need 30Hz or is 10Hz
enough?)


All in all, if I refer to Mark's calculation, my bandwidth requirement
with same frame rate and same image size can be 100 times smaller.
At the end of the day, it seems that you select your hardware based on
your application...

--
Emmanuel







-- D. Peter Siddons Detector Development Group Leader National Synchrotron Light Source Brookhaven National Laboratory Upton, NY 11973

email: [email protected]
Phone: (631) 344-2738

References:
RE: firewire video on RTEMS-4.6.x-MVME5500 Mark Rivers
Frame rate performance (was: firewire video on RTEMS-4.6.x-MVME5500) Emmanuel Mayssat
Re: Frame rate performance (was: firewire video on RTEMS-4.6.x-MVME5500) Emmanuel Mayssat
Re: Frame rate performance D. Peter Siddons
Re: Frame rate performance Emmanuel Mayssat

Navigate by Date:
Prev: RE: ASYN - calling read after interrupt - fix :) Heinrich du Toit
Next: Re: Use of waveform records Touchard Dominique
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  <20072008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: RE: Frame rate performance Mark Rivers
Next: Re: Frame rate performance (was: firewire video on RTEMS-4.6.x-MVME5500) Steve Shoaf
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  <20072008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 10 Nov 2011 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·