Minifloat
In computing, minifloats are floating-point values represented with very few bits. This reduced precision makes them ill-suited for general-purpose numerical calculations, but they are useful for special purposes such as: Computer graphics, where human perception of color and light levels has low precision.