Most HRD software up-rezs the composite images to a 32 bit composite. If my calculations are correct that's 4,294,967,296 colors.
There are bits and there are colors and the two may or may not correlate in any meaningful way. Lets not get color gamut, bit depth/encoding and range all lumped together.
Here's my understanding (and I'm more than open to correction, I am getting my head around all this HDR, Tone Mapping semantics).
The need for 32 bit encoding and floating point math is to have an unlimited set of values for describing what can be a huge number of tones. We all know that an 8-bit per color document in of itself doesn't have less or more dynamic range than a 16-bit per color image. The 16-bit per color document has the POTENTIAL to have more range. That is, if you have more range of tones than you can define with specific values, that's a big problem! So with 0-255, you can have a pixel value of 89 and 90. you can't have a value of 89.5 any more than you can have a value of 89.7. So my understanding is, higher bit depth along with floating point math provides an unlimited set of values to encode what can be a huge number of tones. A 32 bit LDR image is still LDR!
Tone Mapping. In a generic sense, any time you alter the tones of an image, using curves for example on a 24 bit image, you are tone mapping. Sound reasonable?
HDR and the range of the scene versus the capture. I agree that bracketing a scene that the camera can capture in one shot probably should not be called HDR. But playing devils advocate, lets say the scene is 12 stops and your camera can capture that range. But you bracket 2-3 shots and load into your HDR software of choice to tone map an appearance you desire (lets not go into the ugly HDR effects look that make myself and others want to
vomit). We could alter sliders in our raw converter on one capture yes, and get the rendering with one image. But is it possible that bracketing and using a product we prefer, we can alternatively tone map better/faster/easier? Of course that’s not a great route to take if something is moving, you don’t want to go the tripod route etc.
I've been playing with this a bit using just one image (tone mapping). In Lightroom I build a virtual copy, apply two different tone mapping moves but use Enfuse to create one image. First of all, I find Enfuse does a magnificent job of HDR/Tone Mapping, whatever you want to call it with a very natural look. Its also easy to use and cost very little (donationware). I've taken bracketed images into HDR Express, Photoshop's HDR and Photomatix and keep preferring the clean and natural look I get from Enfuse. Plus it works in LR which I love.
While a single capture may have all the tones we want to express, is it unreasonable to use the better HDR/Tone Mapping tools to produce a rendering we wish to express?