No faces of people were in the training data.This was trained on 40 photos that are almost entirely redacted. Most of the dataset was almost entirely a black box with the evidence number in the lower right corner.You can change the case number and slightly alter the subject. Might take 3-5 generations to get close to what you are after. Check prompts on example images for more info.The three prompts that can be used to change the evidence number and (SUBJECT):Completely black redaction:3pst31nr3d4ct3d White page with large black box over most of the page. Text of "EFTA00003238" in the lower right hand corner.Photo outline around black box:3pst31nr3d4ct3d White page with large black box over most of the page. Outline of photo around the black box. Text of "EFTA00003258" in the lower right hand corner.SUBJECT of photo partially visible (ex: hands, lower body):3pst31nr3d4ct3d White page with large black box over most of the page. Photo around the black box of (SUBJECT). Text of "EFTA00003962" in the lower right hand corner.

Files with SHA256: ae0fb8469a4d1b5c582db54389b78597c98c1ee2d94aadc5a3b2c99587e9c7f4