it doesn’t actually know or understand what a finger is. it kinda knows what 2D pictures with people look like, but the pictures it got trained with had fingers in different positions. so when it generates humans, it’s adding a composite of that and it doesn’t count or has an idea about how bones or DNA usually restrict number and position of fingers.
it doesn’t actually know or understand what a finger is. it kinda knows what 2D pictures with people look like, but the pictures it got trained with had fingers in different positions. so when it generates humans, it’s adding a composite of that and it doesn’t count or has an idea about how bones or DNA usually restrict number and position of fingers.