Snow, like titanium dioxide paint, is brilliant white because it’s full of randomly oriented interfaces between transparent media of different refractive indices. This causes entering light rays to execute a low-loss random walk which ultimately comes back out of one side or the other of the metamaterial.
Kerr cells are, as I understand it, normally used to slightly retard the phase of light beams passing through a medium with a strong Kerr effect by applying a few thousand volts across a thin layer of it, thus slightly altering its refractive index.
A much more visible effect of the Kerr effect would result if the Kerr medium were mixed with transparent particles of a different phase, no Kerr effect, and a slightly lower refractive index. By matching the indices of the two materials, the cell would vary between transparency and translucency; when the Kerr effect caused their indices to match perfectly, the phase boundaries would cease to refract or cause total internal reflection, becoming perfectly invisible. This amounts to a sunlight-readable display, like a super-fast high-contrast reflective LCD.
Ideally the two media would have the same permittivity to prevent the electrical field from varying according to the local concentration of Kerr medium, but I don’t know if that’s possible.
In a variant of this approach, instead of particles, the other transparent medium occupies a triangular section of the Kerr cell, while the Kerr medium occupies a different triangular section, and light travels through the Kerr cell lengthwise, entering and exiting at the edge rather than parallel to the electric field. A time-varying electrical field will produce a time-varying refraction angle where the light beam strikes the oblique interface between the two materials, thus deflecting the beam through varying angles, perhaps including total internal reflection (at which point the angle becomes insensitive to further voltage changes). Possibly many different “pixels” can be obtained, each deflecting a different part of the same beam through an independent angle, with bandwidths up into the tens of MHz or more. In this case there is no particular requirement for their indices to be very close. If feasible, this is a better solution to the problem described in Lenticular deflector. Heck, this is probably a product that already exists.
This device as described can also provide amplification; like a MOSFET, its switching action is voltage-controlled and nondissipative, so it can shunt around much larger amounts of power than those needed to control it.