Hi all,

I’m thinking of building a system myself but I have question about the resolution of the camera. I noticed that a majority of people on the forum use 640x480 at 30/60/90fps so does it mean that the application is limited to a resolution of 640x480 as well? But looking at some of the screenshots that the members have uploaded, it looks like the app is running at a higher resolution. Does touchlib automatically scale the captured images to match the app?

I also understand that running a higher resolution at 60fps and above on the web cam will probably cause the comp to lag, thats if the web cam can run at a higher res. optimally in the first place.

Please advise

Melvin

Not sure if Touchlib does the scaling, but its not difficult to map input coordinates to output coordinates.  higher resolutions of camera and more FPS means that there is more image data to process, which is computationally expensive each time to calculate blob location and compare with the last frame.  One thing that the calibration of Touchlib does is work out how much to adjust the raw camera data for any skew of the camera in relation to the screen.  The camera is often not perfectly square to the screen surface.  That may in itself also compensate for difference in size or aspect ratio of the display (it should do).

One thing I’ve been thinking of testing is how acute an angle you can have the camera to the surface.  This allows you to reduce the depth of the project/vision area behind the surface, assuming you can still project the image.  An acute angle would mean you would compress the video resolution in one (or more) plane, which will have some effect on tracking accuracy in that plane (less resolution = more granular movements).

no, app resolution has nothing to do with your camera resolution. It just means that your presses will be more or less accurate. Touchlib finds the weighted center of each finger blob so it’s resolution may be greater than the resolution of the actual camera (it’s subpixel accurate).

Ok THanks. I’m still a little fuzzy about the the whole input/output bit, but I guess I’ll understand it a little better once I try it out.

I run my camera at 320x240 and my apps at 800x600.  Works fine.

Camera res will be important when poeple start developing fudicial/object recognition based tables though.