1. When generating a thumbnail make it so we can use costum instructions for the thumbnails, like: Looking at the camera, Looking sad, A man looking at the camera.
  2. When generating a thumbnail we can chose for the faces to be automatically detected from the video so we dont have to give a picture (this whould allow for the thumbnails to have the actual clothing from the video), and we choose wich face/faces to use in the thumbnail.
  3. Instead of being able to just add in a picture of a face, make it so we can also put other objects or faces we want to be used inside the thumbnails, and then give costum instructions for the ai on how to generate the thumbnail.