I had this idea on the bus the other night so I thought I’d share it in the hope that it might be possible to do.

The basic idea is to take two video streams at the same resolution (e.g. 30 or 60fps) and to combine them into a single video feed in software. If you can manage to find cameras which consistantly and regularly output an image every few miliseconds then it should in theory be possible to stich the feeds together.

There are however a number of difficulties with this:
1. You’d need two webcams of the same model.
2. The webcams need to both reliably output data for a frame at a consistant frequency. 
3. You’d need to be able to start and stop them at slightly different times so that they follow one after another (shouldn’t be too difficult at 30fps but I’m not sure about 60). Which brings me to point 4
4. You’d need to build a small circuit with an led that turns on and off at specific frequencies so that you can sync the cameras.

So it’s not a particularly easy thing to do, but if you had a fast enough cpu then it should be possible…

What do you all think, Could this work?

Well typically you would use 2 cameras to implement a stereo pair (like human eyes) to map 3D space with computer vision.

I really would see much reason to use 2x30FPS… when you could just buy a 60FPS camera…

maybe you could explain your reasoning more…

what about if my camera is to close to the screen and don’t cover the full screen can i just use another camera to cover the rest of the screen and make them work to gather some how?

The feeds can be stitched , but to what avail? The framerate is the same , the only thing that is different is the resolution , using 2 cameras would be incredible for fiducial tracking , but wouldn’t really give any kind of advantage to blob tracking , I’ve been testing with a 320x240@30FPS camera and it’s fast and powerfull enough for decent blob tracking , but looses the blob if I move my finger really quickly over the surface.

Sorry, perhaps I didn’t make myself clear. The basic idea is to take two cameras with the same frame rate, but set them up so that they’re slightly out of sync with each other, ie so they don’t fire at the same time. What that’ll give you is two input streams of 30fps or possibly even 60fps. The important bit, is that the streams could be combined to form an effective stream of 60 or 120fps. e.g.:
Camera 1 Frames: X X X X X X
Camera 2 Frames:  X X X X X
Result:  XXXXXXXXXXX

Call it a thought experiment, but would this work?

Giggs - 24 June 2008 04:04 AM

what about if my camera is to close to the screen and don’t cover the full screen can i just use another camera to cover the rest of the screen and make them work to gather some how?

This is something im interested in as well actually, to have two camera’s to cover the entire screen, can it be done?

Actually, to be more exact, I have got hold of a second hand glass table-top that I am going to use for my coffee table, however as it is 120cm x 40 cm using one projector I am only going to be able to use a small part of the surface as the table isnt that wide. What I was thinking of was to have two projectors project so that the screens are end on end like with a dual monitor arrangement, but with projectors. Am pretty sure this is possible with windows as long as the graphics card is up to it, where the idea falls down however is I would either have to have a camera that can cover the whole length of the table, or have a camera each for the projectors and then have some software merge the two images into one long image.

Is that possible?

Stelph, yes it’s possible and it’s something we’ll be implementing into the tbeta software (multi-cam support).