For example, Filament's documentation at https://google.github.io/filament/Filament.md.html uses the term "dielectrics" when contrasting nonconductors to conductors, which it calls "metallics". And here on stackexchange, https://computergraphics.stackexchange.com/search?page=2&tab=Relevance&q=dielectric also produces many hits for "dielectric". I would have normally expected the word "insulator" in these instances. Does "dielectric" stem from some historical source, or is it the accurate term to use?

  • 1
    $\begingroup$ The first edition of PBRT (2004) used the word dielectric. Being the reference for physically-based shading, it could just be that other PBR frameworks that were developed after PBRT all tried to speak the same language. I'm sure the term dielectric was used way before that but my understanding is that this is the term that stuck. $\endgroup$
    – Hubble
    Sep 12, 2018 at 3:02

2 Answers 2


Strictly speaking, dielectrics are not necessarily insulators. For example, salt water is a reasonable conductor but also a dielectric.

The term "dielectric" tends to show up in discussion of the Fresnel effect—how reflectance and transmittance vary with angle. Dielectric materials (i.e. nonmetals) are contrasted with metallic materials as they have different Fresnel behavior, which can be traced back to how the materials react on a microscopic scale to the electromagnetic field of an incident light wave.

So, for computer graphics purposes, the axis "dielectric/metallic" is more relevant than that of "insulator/conductor", since the former directly affects the appearance of materials.


It is not especially related to graphics but physics, and especially, interaction between electromagnetic waves (like light) and matter, i.e. microphysics of optics.

Metals have free electrons, and thus it's quite a sea of almost free moving charges that are interacting with the EM field. Ideally, it would be totally reflected.

In dielectrics the electrons are not free, but still the atoms and molecules behave as a non-neutral set of central positive charges (the nucleus) and peripheral negatives charges (the electrons), smoothly locked in position by forces (you might see that as springs), thus the dielectric term (or dipolar in simplest configurations). So, the whole thing reacts to EM waves by distorting, and when restoring (+oscillating) also causes an EM wave emission (since moving charges). Note that it is the interference of these direct and reactive fields that makes the "EM field in the material" with characteristic light velocity, and thus tilt the propagation angle at material border (also named "refraction).

Thus, the behavior to light is quite different in both cases.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.