The features will allow users to extend, enhance and reframe images using on-device AI models. Processing typically takes a few seconds.