Converts TRGBFloatImage to TRGBImage. Colors in pixels are simply rounded using Vector3Byte. So such convertion not only kills the floating-point precision in float format but also clamps color components to 0..1.
procedure ScaleColors(const Scale: Single);
Every component (red, green, blue) of every pixel is multiplied by Scale.
procedure ExpColors(const Exp: Single);
Every component (red, green, blue) or every pixel is changed to Power(Value, Exp). So e.g. Exp = 1/2.2 gives commonly used gamma correction.