I've been using a very similar technique for a while now, something I kind of just "figured" out for myself through a lot of experimentation in photoshop. I was using Photomatix a lot to merge 7-9 shot sequences of mostly landscapes, but more often than not was unhappy with the results, especially alignment even though a tripod is always used.
A scene shot in ideal light, early morning or late afternoon, often only needs 2 exposures, 2-3 stops apart. I then take both images into ACR and process each for their respective detail areas, then open both images as tiffs in Photoshop. Go to your "Shadow" image and run through each channel, looking for the one with the most detail in the darkest parts of the image, make a selection based on the chosen channel and then create a new layer off your background using this selection. Duplicate this new layer into your "Highlights" image, turn the visibility of this new layer off, then create a "Shadow Darks" luminosity mask from Tony Kyuper, check out the new "Shadow Darks" channel and maybe up the contrast using a levels adjustment, then I use "Surface Blur" on this channel, most often at 10/30 for my image size, 4368x2912.
Now you can click on the channel to make it active as a selection and apply this selection as a mask to your new layer with the shadow detail, analyze the results, and adjust opacity. I also find that you get much more accurate color in shadow areas from an image exposed just for the shadows, so I also make a duplicate of the "Shadow" image in full and import this into the working image, move it to the top of your layer stack and set the blending mode to "Color". Now use the same or a new luminosity mask for this color layer so you have accurate color in the shadows.
I find this a much better way to extend dynamic range within an image, sure its a lot more work, but you get to know the details of the image as you build it up and you have a lot of options for tweaking each layer as you go.
.
