I have a number of times tried the AI based removal tool. When the person/object crosses higher contrast areas, such as in shadow cast by a building, or a the object/person partially blocks a body of water the AI based removal usually does a poor job.
Often I have a second image taken within a few moments in which the person/object has moved.
It would be very nice if there was a way to point the AI to retreive the image data from the second image instead of attempting to generate the removed area.
Going a step further, multiple source images would be the best solution, think of the ability to take multiple shots of a semi crowded street, and have the AI remove people by using multiple shots as people have moved and being able to peice together the images.
Tim
AdChoices