>>17417
Content-aware fill helps to reduce the text/overlay, but it isn't very smart, it generally just estimates what color pixels it thinks it should use based on what's next to the area it's trying to fill, so I use the clone tool afterwards to correct and blend it with what I think should actually be there
The blonde wig one was a pain in the ass though, I had to manually reconstruct the t-shirt from an entirely different photo lol