JongWon Kim jongwon at netmedia.kjist.ac.kr
Mon Jun 14 13:51:02 CDT 2004


Two years ago (2002), we used your option #2 to develop prototype 3D video
delivery system. At that time, we used stereo camera from Point Gray BumbleBee, 
which generated uncompressed 640x480 color stereo frames (around up to 15fps).
We just muxed synchronous-generated left,right frames together (top/down 
combination?) and sent it over to the network. After it is de-muxed at the other 
end, we have gone througth required processing to match the given stereo 
display environment.



PS: We are planning to integrate an updated version of this stereo 3D video
      program to AGTk in future, after we can wrap up the integration of HDV/DV 
      video that has been showcased at last week AG Retreat.

> -----Original Message-----
> From: owner-ag-tech at mcs.anl.gov 
> [mailto:owner-ag-tech at mcs.anl.gov]On Behalf Of Albert J. Hermann
> Sent: Tuesday, June 15, 2004 2:59 AM
> To: geowalltech at dpc.ucar.edu; ag-tech at mcs.anl.gov
> Cc: darin at ucar.edu; tcoffin at ncsa.uiuc.edu; Christopher Moore
> Subject: Re: [AG-TECH] 3D AG
> Hi,
> A few thoughts/questions about immersive viewing from a newcomer. We 
> have a small project here to investigate low-cost immersive 
> stereo with 
> the AG (or even better, a simple PIG node) in conjunction with our 
> Geowall. After a bit of web-surfing, we came up with a few 
> possibilities:
> Option 1 (single camera solution). Use a singe camera with 
> a/v output, 
> in conjuncion with a beam splitter attachment (e.g. the Canon Digital 
> Rebel SLR in conjunction with the Loreo 3D lens-in-a-cap). 
> This yields a 
> single stereo image (like an old-style stereogram, but in 
> video) which 
> could be sent like any other single-camera video stream. 
> There would be 
> no synchronization problems with multiple cameras (as there is only 
> one), but we would need to find a way to split the video frame on the 
> receiving end, to feed it to the separate projectors of the 
> Geowall (or 
> use the brute force approach, of just shifting one of the 
> projectors to 
> overlap the two images)!
> Option 2 (multiple camera solution) Use a couple of standard digital 
> cameras with a/v output, in conjunction with a synchronizer (e.g. the 
> Sync Shepherd and related devices preciously mentioned on the 
> AG list, 
> or the stereo cameras/synchronizers from Point Grey research), fed to 
> two video capture cards (or perhaps there is a card which 
> accepts input 
> from multiple cameras). This yeilds two nice synchronized pictures to 
> the sender's node, but what happens on the receiving end? I 
> would think 
> there would be synchronization problems, as the two signals 
> may travel 
> differently over the web. One could multiplex the two cameras into  a 
> single video feed, but this again necessitates splitting them 
> up on the 
> receiving end, as in option 1). I would guess with the apporpriate 
> device one could also multiplex input from two cameras into a 
> frame-flipped stereo signal (or perhaps single cameras exist which 
> produce this), but could the receiving node correctly unravel a 
> frame-flipped video sent over the web?
> I could well be missing something fundamental about the AG, 
> which makes 
> this process easier or harder than it appears. Any thoughts 
> appreciated. 
> Thanks!
> -Al Hermann
> Darin Oman wrote:
> >This is way cool! We finally checked it out. The cameras are 
> a little misaligned vertically, but I was able to adjust the 
> window accordingly. One thing that might make it difficult 
> for practical AG purposes is that the cameras are a little 
> out of sync, so when there was a lot of movement, it appeared 
> very choppy. I'm not sure how you could adjust that. We'll be 
> playing around with this idea ourselves. One thing I'd like 
> to try is messing with the focal point to see if you could 
> get the image to come out at you rather than away from you.
> >
> >Cool stuff!
> >
> >Darin
> >
> >
> >On 2/9/2004 at 11:25 AM Tom Coffin wrote:
> >
> >  
> >
> >>hey!
> >>
> >>A couple of weeks ago the HASTAC folks were meeting
> >>here at ACCESS and I was demonstrating for them the 
> >>Accessgrid and Geowall technologies. During a conversation
> >>at the geowall, David De Roure from University of Southampton
> >>asked me if anyone had done the accessgrid on the geowall
> >>and I replied not to my knowledge but it would be easy to
> >>setup.
> >>
> >>Well, time went on but I got an email from David asking
> >>if I had setup the 3d ag. I hadn't but it got me to try it
> >>and in under 10 minutes had it up and running.
> >>
> >>Basically, took two cameras pplaced the closely side by side
> >>and pushed their inputs into an accessgrid node. Then installed
> >>the ag1.x pig software on our geowall machine and opened
> >>vic inot our ACCESS-DC institutional venue. Then just opened
> >>the two streams. Placed the right image to the right projector
> >>and the left image to the left projector and bingo! 3D AG!!!
> >>
> >>I have the two images up in the ACCESS-DC Institutional venue
> >>and will leave up for a while. Feel free to drop in with
> >>and check it out on your Geowall.
> >>
> >>
> >>
> >>___________________________________________________________
> >>Tom Coffin .......................... tcoffin at ncsa.uiuc.edu
> >>    
> >>
> >
> >
> >
> >  
> >
> -- 
> --------------------------------------------------------
> Albert J. Hermann
> Pacific Marine Environmental Laboratory
> 7600 Sand Point Way NE, Seattle, WA 98115
> e-mail: Albert.J.Hermann at noaa.gov
> ph:  (206)-526-6495 fax: (206)-526-6485
> WWW: http://www.pmel.noaa.gov/~hermann/
> --------------------------------------------------------

More information about the ag-tech mailing list