h a l f b a k e r y
Is it soup yet?
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
On the rare occasion that I have used PC based webcam/chat software, I have encountered an issue that I find distracting. The person on the other end of the conversation is always looking someplace other than in to the camera making it seem to me as if they aren't paying attention. Since my camera is
on top of my monitor, I must appear to be looking down instead of directly at the person on the other end.
This thought was inspired partly by a recent online chat with my niece who was using her newly acquired web cam and partly by a "limitations of electronic communications" conversation I recently had with an e-quaintance.
It seems that it might be possible to place two cameras, one on either side of the monitor (probably vertically centered), and combine the two images to make a single 2 dimensional image that appears as though I am looking directly in to the lens when I am actually looking at the middle of my screen.
It would probably be necessary for the image processor to know the relative angles and distance between the 2 cameras to figure out the convergence point.
May take too much processor power to work in real time.
May not be possible to come up with a non-artificial looking image.
I have little experience with this subject. I only have a webcam because the net cost was $5.00. I rarely use the video support available with some instant messenger clients.
With regards to my knowledge of digital image processing - on a scale of 1 to 10 - I'd have to do some serious studying to achieve a 1.
I couldn't even figure out sufficient key words to use a search engine to find similar or related ideas. Might be baked.
Records imagery using the monitor itself as an input device. [Shz, Oct 04 2004, last modified Oct 05 2004]
Apple Patent For Camera-Screen
Thousands of optical sensors packed in with the liquid-crystals will create a screen that sees you, probably from multiple angles at once. [ironfroggy, Jun 01 2006]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
||I've seen 3 camera extraction, with great results, but certainly not in real-time.
||Or, for the retro-mechanical approach, use a teleprompter-like setup: your scan-reversed monitor reflected off of a 45° angled, first-surface 50% (or so) silvered mirror, behind which the camera is placed. As you are looking at the righted reflection of your monitor in the mirror you will also be staring directly down the lens of the camera.
||I remember reading somewhere that you need 7 "cameras" or points of reference to map a vector in 3-dimensions accurately. Like when they do motion capture for 3D animated models.
||Oh you poor electronic age people. Go back to proper optics
and read up about offset/shift lenses. You just need to
emulate this electronically - Bill Gates has enough money to
do this if he could be bothered! A lens forms a much bigger
potential image than is captured on a small piece of
film/light cells so, effectively, you point the lens at something
else (e.g. just above your head as you work on your
computer), but position your capture area in the periphery of
the potential image rather than slap bang in the middle (so
the bit of the image with your face in it is the bit you record).
This is how tall buildings and mirrors are photographed. Go
and look in a basic photography book if you don't believe
||Sorry, Ruth. Although an shift/tilt lens would straighten out the perspective you would still not make eye contact with the camera.
||This sounds feasable, after all we're not demanding pin-sharp image quality (yet). Given the power of moden CPUs the online number crunching should be possible.
||For half's idea to work your head would have to be positioned at exactly the right point horizontally, and vertically aligned, at exactly the right distance. And you would appear to be partially crossed eyed.
||"It would probably be necessary for the image processor to know the relative angles and distance between the 2 cameras to figure out the convergence point." Could possibly deal with the horizontal position. I envisioned this as possibly being a dynamic process, akin to human eyes changing angle to focus on objects of various distances.
||Hm...I'll admit that I've thought of many reasons why this wouldn't be worth doing and possibly it's not possible, but apparently I missed something. Why would vertical alignment be critical? That may be why the 3rd camera is required according to [bristolz]' annotation.
||I meant your head has to be vertically aligned, as opposed to tilted or rotated sideways.
||Are you suggesting that the cameras have small motors on them to automatically aim at the persons head?