[mpeg2] kfir latency revisited

[Bram Stolk <bram@sara.nl>]

Hello,


I am trying to determine the latency of the BMK kfir card.
Rob Morris mentioned previously on this list, that the latency
is at least as long as the time between two I frames.

With GOPChop, I see that by default, the card generates GOPs
of 15 frames, like this:

IBBPBBPBBPBBPBB

So I expect a 15 frame delay.

However, experiments suggest a shorter delay.
What I did was modify the mpeg2dec.c sources to use a very small
buffer. Then I start mpeg2dec to read directly from /dev/video0.

The delay I see is approximately one third of a second.

However, as the stream is composed of 25 frames / second, and
the GOP size is 15, I would expect a delay of 15/25 = 0.6 sec.

Are the I/B/P frames 'half-pictures' instead of full pictures
consisting of only even or uneven scanlines?
This would explain the 1/3rd sec latency.

Also, I cannot get mplayer to read directly from /dev/video0
as mplayer first seeks to EOF to check filesize, which is not
appropriate for a device node.

Thanks,

   Bram Stolk


-- 
------------------------------------------------------------------------------
 Bram Stolk, VR Specialist.
 SARA Academic Computing Services Amsterdam, PO Box 94613, 1090 GP  AMSTERDAM
 email: bram@sara.nl   Phone +31-20-5923059  Fax +31-20-6683167

 "I heard if you play the NT-4.0-CD backwards, you get a satanic message."
 "Thats nothing, if you play it forward, it installs NT-4.0"
------------------------------------------------------------------------------