i spent some time trying to make a picture of somebody crying while being massively indoctrinated to shut up and forget, to print to use to help communication with therapists, but it took too long and self-applied, it’s pretty hard (to complete and feel and stuff). i tried to use ai image generation to make it more supported by the patterns i guess? but it took so many cycles of trial it didn’t seem good. thoughts: - embeddings-informed image generation might be faster - free, open services might be more private and safe -> handmaking images seems clearly much safer and more reliable despite any feelings to the contrary - getting experienced with image tooling would also help, making specific practice images. this might work with handmaking too.