A feature that would be enormously helpful to users creating HDRs for Image Based Lighting in 3d software would be an option to 'normalize' an HDR file's output to have all values scaled to be between 0-1.
It is actually at the top of my TODO list
I just wanted to add a +1 for this request, although I don't think simply scaling the output to 0-1 is a great option.
My current work around is to use:
oiiotool --cmul %s luminance_hdr_output.exr
to multiply the output HDR image by a constant scale factor determined from the luminance of a reference 18% gray object in the scene. That enables normalization to the convention used by the the spi-vfx OpenColorIO profile (and others): 0.18 == 18 % gray, 1.0 == 100 % diffuse white, 1.0 < Specular highlight or emmisive.
Being able to specify a scale factor directly in Luminance would be very convenient. Another reasonable, and perhaps more user friendly, default would be to let the user set one of the input images as a hero image and normalize the output to match the hero's exposure assuming the hero was well exposed and an output middle gray value of 0.18.