

Hi,
I was wondering if I could train a neural network to solve a prediction problem: I have a series of bitmaps (photographs basically) covering a few angles around a 3D object (the input data). I would like the neural network to predict the projection of the object from ANY angle around it (out of 360 degrees  i.e only a spin around the object). I was wondering if this is possible (even if it's a slow network it'll do).
If so, then what network architecture and training algorithm should I use to train the network? (I suppose that the number of required neurons will increase with image size, so let's say I have a 256 x 256 bitmap).
Any help will be appreciated. Thanks! 