[AG-TECH] Re: AG video node using berkley mjpeg work

[Lawrence A. Rowe]

Bob Riddle wrote:
> 
> A few more observations:
> 
> on the Win2k Display machine, you can't just slide the openmash vic.exe
> (and vic, mash.exe) in teh /ag/agapps/bin directory, even if you rename
> it to ddvic.exe.  When the DRM manager starts, it expects to find "the"
> ddvic" in that directory, if not (or if renamed vic.exe) a message box
> is prodcued telling you its "can't load ddvic".  The same thing happens
> when you move through the rooms via the web browser.
> 
> On the linux machine used for video capture, I'm trying to understand
> better what vrm-eventlistener does upon startup.  It appears to use a
> "parsed sdp" stream to get the mcast address, but the port isn't
> mentioned in the stdout messages.  Also, it starts vic with a user
> script file "-u /tmp/vic-startup.0" which doesn't seem to hang around
> long enough for me to peek at it.  Anybody know what this script looks
> like?  I'm trying to figure out why I have to start the mjpeg vserver
> before the vrm-eventlistener.
> 
> Anyway, Matthew indicates in his mjpeg pages that vic can be built using
> the lml33 mjpeg drive.  In fact, Larry Rowe had sent a note some time
> ago indicating this might be included in future openmash "releases".  So
> my current thought is to try and get a linux version of the openmash vic
> slide in place on the AG linux video capture machine and see if it
> resolves the startup issue by using more "normal" AG video capture
> software.
---

That's note quite correct.  The LML33 capture interface in OpenMash only
does raw video capture - that is, it uses the LML33 to capture the frame
then hands it to the normal software encoding routines.  You can encode
the captured video as mjpeg and it will be similar to RTPtv streams, but
probably not identical -- i.e., I don't think we have restart markers in
the OpenMash code.

Matt, will RTPtv vclient replay mjpeg streams without restart markers?
	Larry
-- 
Professor Lawrence A. Rowe          Internet:  Rowe at BMRC.Berkeley.EDU
Computer Science Division - EECS       Phone: 510-642-5117
University of California, Berkeley       Fax: 510-642-5615
Berkeley, CA 94720-1776            URL: http://bmrc.berkeley.edu/~larry