Hi, I’m working on a way to automate simple 2D projection mappings using openCV.
So far I’ve been able to use ofxCV to detect contours and quads in a given image/video and adjust the content to that quad. Everything happens on the screen, so far so good.
When I move to using a projector and a webcam I run into the camera-projector issue. I need an image of what the projector ‘sees’ so I can process it and feed it to the contour finder. Madmapper has the ‘Spatial Scanner’ feature that does this using a gray code pattern that is projected onto the target surfaces and then using a camera it processes the images to build an image of what the projector sees. You can see how it works HERE.
I’ve looked into ofxGraycode by @elliotwoods but the resulting image after decoding the gray code is a depth map and not what the projector ‘sees’.
So…the workflow should look something like this:
Set up webcam and projector.
Project the gray code pattern.
Decode the images that the webcam captures and get what the projector ‘sees’ ----> Here’s where I need the help
Process the image that the projector ‘sees’ to get contours, quads, etc.
Fit the content in the contours.
Any help will be appreciated…or maybe there’s a different approach that I’m not aware of…check out this one for example…