This is a very interesting circuit. I've used similar, for very fast (10nanosecond pulses) amplification with 2N3904/3906, where I needed better dynamic response than opamps could provide and still work within 0/5volt rails (this was many years ago). This circuit is actually 1/2 of a current mode opamp, and the emitter of the NPN is the summing node.
The input circuit, especially with that diode, looks like a sync-tip clamping circuit, a peak-detector, that controls the biasing of the entire amplifier. Thus the black and the white levels depend on how you set P1.
The feedback circuit is P2, and setting the pot wiper in the middle should provide a gain of TWO. This gain of TWO is needed to overcome the 2:1 voltage division of the 75ohm Rout and the termination resistor, also 75 ohms, at end of the video coax cable.
Now ---- why do your modifications not work?
Knowing this circuit rather well, my first thought is "That 5Kohm pot P2 is the ONLY DC pulldown. That will not work."
The 75 ohms is also the DC_pulldown (actually 75 + 75 = 150 ohms).
I suggest you remove the output capacitors.
The extra 14 milliamp may overload your +5 power supply. Probably not. But check that.