Peter, I believe the Xtherm and Xtherm Max are similar when layered with a CCF pad. Without any sort of insulating materials (just baffles and air,) they have a hard time changing the physics of the pads.
Using a Zrest, (mylar? coated) will block any IR anyway. Unfortunately, any that does get through will get trapped between the Xtherm and the Zrest eventually warming the Xtherm’s upper layer and not reflecting this back to you, directly. The standard NeoAir does this, too. Again, theory says that more baffles is better. A third & forth layer of mylar between the layers, added Xtherm style, will add more than increasing the number of horizontal baffles, though.
Similar to CCF pads, more bubbles per inch of thickness is warmer for identical substrates. So, at the base level, a standard NeoAir initially looks to be equivalent, but the smaller baffles in the triangular shape will allow greater heat stratification. Except at the very apex of each triangular shaped baffle, this is roughly the same as adding a second internal baffle to the Xtherm between all the baffling present top & bottom. (And incidentally increasing strength and easing manufacturing methodology.) In this case, due to the shape, more baffles is also supplemented by an effective additional baffle between the top & bottom, and, is cheap/easy to produce. The fluff simply adds a little more insulation and reduces the “plasticy/sticky” feeling of the original NeoAir.Discounting edge bleed, the XTherm will be a few degrees warmer.
Edge bleed becomes the problem with the larger sized Xtherm at low temps. The Xtherm appears to be optimized for the middle latitudes of the world and not extreme low temps due mostly to the edge bleed. Not strictly true, but you can think of the edges as where the most unheated area of the pad lie. All layers are joined at the edges increasing general conductivity. And the air baffles are all joined there with no insulating value except what the plastic sheets provide. There are no air pockets at the edges to provide any insulation. Every baffle is joined to the edge with the horizontal construction. (I believe they allow internal venting at the edges to simply manufacturing but never took one apart to see.) Since the Xtherm/Max is wider than the typical sleeper, this means any heat loss will connect every baffle at the edges. Edge bleed becomes an important consideration and is not generally considered in Rvalue testing (usually done at each surface to measure the restriction on heat flow.) R value testing is usually static.
Any outside air flow will change this a lot, since convection becomes pretty much unrestricted at the edges. Any movement will cause the cold and warm air to mix. So: sleeping in a tent vs sleeping under a tarp will likely be a different number. Pressure from the sleeper can collapse the internal baffling rendering them ineffective (side sleepers vs back sleepers for example.) Covered, the Xtherm looks to be an excellent choice. Uncovered, it has the same edge bleed as the original and wide pad offerings from Therm-a-Rest. Edge bleed nullifies any extreme cold use without some sort of CCF pad over the whole thing. But, the Xtherm is better at raw R values. How much is difficult to say, it depends on your bag and comfort levels of each sleeper. Again, only the first baffle layer will actually reflect any IR back at you. That they chose to picture the top layer reflecting is consistent with my statement that the best use of multiple layers will have the best insulator on top and as Bob was saying.
If I had my druthers, I would select a pad with a full vertical tube on each side NOT connected to the smaller horizontal baffles on the inside of the pad, and, optimized to insulate against edge bleed (perhaps at a 45 degree angle.) Blow it up firm. This will prevent baffle collapse. And cover completely with a layer of CCF. The Xtherm works well, but could be considerably better for a slight increase in weight. It would increase the manufacturing difficulty a lot, though. Hell, “best” is the American way. To pay an extra $25 because of the improvement is likely feasible, though.