I often read in forums people making HDR with only one RAW picture. They 'develop' from this RAW a first picture with 0EV correction, a second with +2EV and a third with -2EV. Then, they make HDR from this 3 pictures.
For me, with that solution, they only use the 12bits of the RAW picture for the HDR, nothing more. It is not really HDR, as it does not exeed the sensor limits.
Am I wrong?
Canon 20D + 17-40/f4 L USM + 70-200/f4 L USM + 50/f1.4 USM
Merlin/Orion panohead + Papywizard on Nokia N800 and HP TC-1100
However, it is an excellent method sometime to balance your images through "exposure blending". I've made this for some images and it works fine, as RAW format has more color information than the same image in JPG. It is simple to extract a "dark" and a "light" image if your RAW image has a large lighting range.