I don't know if you have the option to, but you could try just throwing out the most significant bits of the data and mapping the 8 least significant bits (or maybe drop one or two lsb for noise and pick the rest). Basically, modern thermal cores should be outputting more than 8 bits of data - 14 bits is pretty standard, but for 24 bit color, for example, you really only have 256 greys to play with. Now more modern displays may give you 10 bits of color and then the same number of greys, but that's still quite a bit shy of the dynamic range of a thermal camera, so they scale the extra bits into different color channels, so we can better visualize the full dynamic range with our eyes that see more than just black and white.
I've played around with some manual pixel depth configurations in some digital capture devices and you get a very neat banding effect somewhat similar to a rainbow palette when you look at 14 bit data in a mode looking at fewer bits, because you see the ramping up of the thermal gradient in the frame, and then it suddenly resets as those more significant but now hidden bits turn over. Not great for a normal looking image, but great for seeing the detail of the dynamic range in an image.