If I was to do it I would take every digital frame and then take each pixel from top left to bottom right and make one long chain. Either that or average x neighboring pixels to fit the size vertically. I have no idea how they did this specifically though.