I don't know how interesting this will be to anyone, as most of you are already pretty familiar with this process...but I made a convenient little action for it, so anyone who wants to save themselves a bit of time should feel free to download it!
For those of you not familiar with this NR method,it can be summed up like this: you take two or more identical-as-possible high-ISO shots of the same subject, layer them together in Photoshop, and use Smart Object blending to average away a good deal of their noise (anything that doesn't exist in all of the images disappears). It isn't the most convenient NR method, as it requires several shots (all of the ghosting, etc. rules of HDR apply), but as long as what you're shooting is fairly static, it is very effective and, most importantly, almost completely non-destructive. It's horrible for concerts and action shots, but perfect for low-light landscapes and portraits (provided your subject can sit still). And, of course, the faster your camera's FPS, the better. 7D users should have few problems pulling it off!
Here's an example, made of three combined ISO 3200 shots with shadows pushed just a touch. Before (one shot) on the left, after on the right:
|IMAGE IS A REDIRECT OR MISSING!|
HTTP response: NOT FOUND | MIME changed to 'image/png'
As you can see, what it has done is basically knock the noise down at least a full stop, if not more. It obviously hasn't magically eliminated all noise, but in the very least it's allowed us to be much, much less severe with any subsequent traditional NR.
Anyway, instructions are simple:
1. Open your two or more shots in Photoshop
2. Copy and paste all of them into one file, leaving each as a separate layer
3. Select all of said layers
4. Run the action! (I have it set to F12, but you can change that)
5. Unless your shots are PERFECTLY identical, you will most likely have to crop a tiny bit of the edges off to compensate for layer alignment.
6. You won't lose any detail (in fact in most cases you'll gain some), but you may lose just a tad of perceived sharpness. So keep your USM at the ready.
For reference, what this does is auto-align the layers, then combine them into a Smart Object, then set the smart object's blending mode to Mean. Not very complicated!
Note: this was made in Photoshop CS5 on a Mac. I have no idea how it'll play in anything else, so please don't shoot me if it doesn't work for you. In any case, it's not all that difficult to make yourself!