I can understand how they visualised the photon packet passing up the coke bottle - it requires a camera with an exceptionally short sample time (shutter speed) and a light source (laser) able to produce exceptionally short light pulses. Presuming they have not cheated and used a huge model of a coke bottle, the pulse in the video appears to be about 3cm long, so the camera sample time plus the laser pulse time must be about 0.1 nanoseconds. Don't ask me how they do that - in my hay day I was designing with logic gates which took a few nanoseconds to respond to an input (just one gate), so how you make a whole camera sensor turn on and off in maybe 0.05 ns is well beyond my knowledge.
What you see is the scatter from the medium in the bottle (dusty water), but at very low light levels. They then have to do what astrophotographers have done since digital imaging became available - many, many perfectly synchronised runs (each run only capturing a few photons, but in this case synchronised to within 0.01 ns - 10 picoseconds!) and add them together.
Computing the 3D shape of an object out of direct line of sight from the incredibly subtle variations in shadow... that's another matter!
A camera to see through solid objects? Well, we (physics students) are told that there is an infinitesimal (but not zero) chance of one object (photon) tunnelling through another object (barrier) by quantum probability (Schrodinger's wave equation and all that). The lecturer traditionally bounces a tennis ball off the theatre wall repeatedly, then calculates how likely it is that the ball will pass through instead of bouncing off. He would drop dead of shock if it actually did.