Why assign a participating medium to a sensor?

[Disgruntoad]

Based on the first few paragraphs of the documentation for participating media (https://mitsuba.readthedocs.io/en/latest/src/generated/plugins_media.html), "When a medium permeates a volume of space (e.g. fog) that includes sensors, it is important to assign the medium to them. "

Can someone explain why we need to include the sensor? Do we also need to create a cube to enclose the scene? What exactly does assigning a medium to the sensor do? Thanks!

Secondary question, I've seen some mediums use the assignment <bsdf type="null"/>. Is this important to explicitly call (and what does it do)? If we do not do this, what exactly occurs?

[njroussel]

Hi @Disgruntoad

Can someone explain why we need to include the sensor?

The sensor's medium is used in the integrator to know if the ray starts inside a medium or not (relevant parts of code: 1, 2). It's important to know if the ray's origin is already in a medium, as it will tell us whether or not we need to handle medium interactions from the very beginning of its light path.

Do we also need to create a cube to enclose the scene?

No, usually not, but it is safe to do so. Mitsuba tracks which medium a ray currently is in by tracking medium transitions when hitting a shape - there's isn't a global data structure that allows queries like "tell me which medium this 3d point is in". Hence why we also need to tell it its original medium. So, if your ray is always inside the medium you don't need to enclose it in a shape, as the shape's sole purpose is to define a transition.

What exactly does assigning a medium to the sensor do? Thanks!

I think my first answer covers this.

Secondary question, I've seen some mediums use the assignment Is this important to explicitly call (and what does it do)? If we do not do this, what exactly occurs?

I'm not quite sure what you're referring to. Could you point out some piece of code?

[Disgruntoad]

@njroussel Thanks for the detailed answers!

Mitsuba tracks which medium a ray currently is in by tracking medium transitions when hitting a shape

Just to clarify this point, if I have declared a fog medium in the sensor, then include smooth dielectric materials in the scene, will the rays behave like they're in a medium after leaving the glass?

Looks like I accidentally forgot to include part of the original question. I've seen

<bsdf type="null"/>

in some of the demo samples mitsuba provides. This is typically used for either mediums or emitters.... I'm not entirely sure what it does or when it is proper to use it.