More specific question?
In "matrix operation", your data is in a matrix before being displayed in a window, so you should probably use this data in the first place.
If you use OpenGL, you might get you data "after the window". Is it the case?
I'm using OpenGL, which is basically rendering to shapes what this external is outputting to jit.window.
Since I can't alter the code for the external, I have to stick to the video signal coming out "after the window", as you put it. I know this is possible and very basic so if you can help me, it would be very nice.
Two more things, if only a predetermined kind of data is allowed to enter jit.window, then it would be theoretically be possible to alter this signal "before" coming to jit.window? (e.g. a Matrix).
And one last question, regarding OpenGL. Is it possible to render multiple shapes on a single jit.window, make them interact and react to the camera, a la Raycaster?
Hmm, it's still not very clear. You're using an external that outputs to jit.window? In that case, the external should have at least a context name. You could then, instead of rendering to a window, render to a matrix. An example of this is given in the help for [jit.gl.render].
Your second question is kind of similar to the first. I'm not so sure what you mean by 'after and before the window', as all GL rendering is done on the videocard, but indeed it's possible to get all this back into a matrix, at the loss of some speed.
Last question: yes, of course you can draw multiple shapes in one window. How do you suppose 3D games work? You'll have to make the interaction yourself (with 3D coordinates interacting), but it's very possible.
Is it still possible that GL rendering to a matrix is in the works? The tutorials mention that this might be coming, but they make a reference to Mac OS 9 in the same sentence, so I have no idea if this is never going to happen, or might be something we'll see in Jitter 2.0.