What if they would use more then one sensor and lens? A cluster of crappy cameras, and then compute a good result?