I can finally post this now. A somewhat solipsistic view of our progress with reading scrolls. Brought to you by Mathematica and ffmpeg.

what does "color in the computer screen some more" mean? also what is this one view? does any of the raw unprocessed data have characters that look that clean?

With a marker.

I love that the winning combo ended up being: machine learning + coloring with markers

Curious where FFMPEG comes into play with your methods?!

Just how I made the video.

probably a massive oversimplification but are you basically making a distinction between burnt(paper) and burnt(other) and then analyzing the burnt(other) in cross-reference to the cultural-in-question’s alphabet?

this is absolutely insane

Is there an open source database of untranslated scrolls?

this is so unimaginably cool. congrats

You need fancy tools to find a pie, but I just have to look in my fridge.