Look, if you have enough memory and enough power within your cam, you could develop HDR stacking soft to transfer raw memory into alternative memory, using HDR algorythm (upon Ev), and then save this alternate memory into OpenHDR formated file.
Or even do it in small parts from disk, as "raw sum" and "raw average" do. But the camera is poorly suited to this. It's slow, and the hacked environment CHDK runs in prevents you from taking advantage of existing software libraries.
There's slightly more justification for doing this on a wireless device like a smartphone (leaving aside the fact the sensors and optics are garbage), since you can publish the picture without ever touching a PC. They also have much more memory and CPU power. Even so, a cloud service might be more appropriate than doing it on the phone.
