By using a shared exponent, the RGBE format gains some of the advantages of floating-point values whilst using less than the 32 or 16 bits per color component that would be needed for single precision or half-precision data in the IEEE floating-point format, and with a higher dynamic range than half-precision.
Greg Ward provides code to handle RGBE files in his Radiance renderer.
RGBM is a format with the exponent replaced with a shared multiplier, while RGBD stores a divider instead.
These formats lack the dynamic range of RGBE and logLUV, but are more amenable to a naive approach of linear interpolation on each component.
[4] A wider variety of color formats take the more conventional route of storing separate floating-point numbers.