The following functionality could significantly improve the results and eliminate previous errors. The option of selecting parts of the image (e.g. with a lasso or brush) and then defining them more precisely by reference. For example, text that is not recognized correctly could then be specified by text input. Or faces could be optimized with a reference image. Or materials, e.g. mark hair and then say that it is hair. Or simply not to perform any enhancement at all in the selected area, so that it remains in its original state. e.g. small faces in the background, which are often ugly. I’m sure this would be a huge improvement for the App.
