Photos have optical distortion, you know, and AI is the wrong tool for extensively manipulating pixel data. Or in short: Your workflow proabbly doesn't make much sense. That is not to say that it would/ should be impossible, but you will need to spend more time with the program and learn some advanced techniques to produce freeform deformations, use wrap objects, clipping masks and so on. It really comes down to experience and knowing this stuff, which would also be true if you were using Photoshop. You simply can't expect to download a complex software and tackle a reasonably complicated workflow right away.
Mylenium