The author has taken a purely emprirical approach and not attempted to model the emissions or localize them to a particular point, so we don't know. You would expect differential signalling to be low loss, but perhaps the termination is bad or the shielding around the connector? Or it could even be being radiated from the power (VCC/GND) of one of the chips on either end.
There’s some kind of sync mechanism in HDMI, too, but I don’t recall what it is. Or possibly data islanding?
Though it's entirely possible that Pi leaks HDMI into the ether. Someone with a Pi4 and a RTL-SDR could easily verify it: pixel clock frequency of a 1024x768 signal is around 50 MHz.
It'll be some pseudo-random mapping, not "greyscale", but it'll still be readable.
Look up the wikipedia page for 'ECB mode' for a nice illustration.
How does this work? Specifically:
> Maybe, if we use the strong pulses for synchronisation and plot the amplitude of that noise as a two-dimensional picture, we could see something?
From surface level knowledge, HDMI uses Transition Minimised Differential Signalling (TMDS). A direct plotting of noise from that transmission line should not provide even the low quality greyscale image repro shown. This may just be a titling issue but I'm assuming that based on the info shown the RF source must be the display electronics rather than HDMI line? What am I missing?