-
#Idea: A #MachineLearning system which uses #ComputerVision to estimate the "quality" of an image from a human perspective (how readable the text, sharp the edges, etc.) then use that information to compress the image as much as possible with the minimal dip in observed quality.
-
Motivation here is to optimize my tasteless self out of @jaffathecake's #Squoosh compression loop. youtu.be/F1kYBnY6mwg?t=1421