I'm currently working on a flash application that will be used to display architectural interior renderings with various interactive features. One of the features is to be able to control the strength of different light sources. Pretty much trying to replicate a simple version of the Multilight (http://www.maxwellrender.com/maxwell_firsts.html) feature in Maxwell Render.
I have bunch of HDR images, each exported with a single lightsource active. Now combining these images is just a simple matter of compositing the layers in Nuke or Photoshop, using over and linear dodge(add) respectively. This produces the same result as exporting an image with all the lightsources active.
Flash however doesn't support 32bit images, so in order to do the same compositing in flash, the images must first be converted to 8bit. Unfortunately, compositing the 8bit images using linear dodge(add) in photoshop or flash doesn't produce the same image at all. So I assume in order to get the correct result you need the data that is being stripped away when converting to 8bit.
The question is
Is there a way to convert the 32bit images to 8bit, so I can correctly add them together?