Comment by nunodonato

wow, what kind of stuff does that script do? I've seen non-vision models analyze images, but mostly histograms, color averages etc. This one seems to actually understand the image itself and reproduce the layout, very impressive