at least the model we used was behaving like this. It was making objects larger than it visible as a line and smaller visible as a whole (like coloring fill in

at least the model we used was behaving like this. It was making objects larger than it visible as a line and smaller visible as a whole (like coloring fill in MS Paint). Objects that fit in the eye were colored and larger were not. However this was a neural network model (shown in excel as a matrix) - I wonder where did he get it from. When I wrote the above, I tried to find it but couldn't