Question

Just getting started with OpenFrameworks and I’m trying to do something that should be simple : test the colour of the pixel at a particular point on the screen.

I find there’s no nice way to do this in openFrameworks, but I can drop down into openGL and glReadPixels. However, I’m having a lot of trouble with it.

Based on http://www.opengl.org/sdk/docs/man/xhtml/glReadPixels.xml I started off trying to do this:

glReadPixels(x,y, 1,1, GL_RGB, GL_INT, &an_int);

I figured that as I was checking the value of a single pixel (width and height are 1) and giving it a GL_INT as type GL_RGB as format, a single pixel should take up a single int (4 bytes) Hence I passed a pointer to an int as the data argument.

However, the first thing I noticed was that glReadPixels seemed to be clobbering some other local variables in my function, so I changed to making an array of 10 ints and now pass that. This has stopped any weird side-effect, but I still have no idea how to interpret what it’s returning.

So … what’s the right combination of format and type arguments that I should be passing to safely get something that can easily be unpacked into its RGB values? (Note that I’m doing this through openFrameworks so I’m not explicitly setting up openGL myself. I guess I’m just getting the openFramework / openGL defaults. The only bit of configuration I know I’m doing is NOT setting up alpha-blending, which I believe means that pixels are represented by 3 bytes (R,G,B but no alpha)). I assume that GL_RGB is the format that corresponds to this.

Answer 1

If you do so, you need three int: one for R, one for G, one for B. If think you should use:

unsigned char RGB[3];
glReadPixels(x,y, 1,1, GL_RGB, GL_UNSIGNED_BYTE, rgb);