Near-IR video cameras aren't much different from visible-light ones. The one in this clip looks pretty sophisticated by comparison.
However I have to agree that the demonstration is nowhere near quantitative enough to infer much about absorption by CO2 of thermal radiation from Earth's surface. A far more accurate method is to calculate it line-by-line from theHITRAN line spectra tables.
However mere absorption of surface radiation is only about 6% of the impact of CO2 on global warming even in the no-feedback case. This is because what heats the Earth is reduction in outgoing longwave radiation (OLR) from the top of the atmosphere (TOA). Only 6% of that radiation is emitted by the surface, the rest is radiation from clouds and greenhouse gases in the atmosphere.
Clouds are not water vapor but droplets, which unlike water vapor but like the surface are much closer to being black body radiators. Although there is somewhat less CO2 above the clouds than above the surface (the difference being the amount of CO2 between the clouds and the surface), it's quite enough to absorb the same bands emitted by the clouds as those emitted by the surface.
Radiation from the atmosphere's greenhouse gases is narrow-band, even at sea level but increasingly so at higher altitudes as the effect of pressure-broadening decreases. Every greenhouse gas emits its own set of lines, and absorbs the same again, so there's a lot of emitting and absorbing going on in the atmosphere.
Looking down from above the atmosphere, a thermal imaging camera sees only the "top layer" of all this radiation. This layer is not sharply defined but rather is a separatephotosphere for each wavelength of IR. To quote the Wikipedia article, "The photosphere of an astronomical object is the region from which externally received light originates." Wavelengths that are absorbed more strongly create more opaque and therefore higher-altitude photospheres. The further below the photosphere, the lower the probability that a photon from that depth will escape to space. The probability is nonzero however no matter how deep, whence the indistinctness of each photosphere.
What increasing any greenhouse gas does is to make it more opaque, thereby raising the altitude of the photosphere associated with each wavelength at which that gas absorbs and emits. The higher you go the colder, namely 10 C/km for dry air (the Dry Adiabatic Lapse Rate or DALR) all the way down to 5 C/km for saturated air (the Moist Adiabatic Lapse Rate or MALR) when very warm. Hence a higher photosphere is colder. And since radiation follows the Stefan-Boltzmann law, the amount of radiation falls off as the 4th power of this decreasing temperature.
Higher temperatures raise the water vapor in the atmosphere. Hence heating the atmosphere by increasing the CO2 will increase water vapor, another greenhouse gas, which in turns heats the atmosphere even more. This vicious cycle is called a positive feedback, and is believed to add considerably to the basic no-feedback greenhouse effect attributable to CO2.
Richard Feynman said of quantum mechanics that if you think you understand it then you don't. The greenhouse effect is not quite that bad, but it runs a close second. John Nielson-Gammon has offered "The Best Ever Description of the Atmospheric Greenhouse Effect". I don't know if my account above is as good, but it's only half the length.
-Max