m_distribution properties

[powelld]

Hello,

I am wondering if someone could provide some information on the two distribution types for the m_distribution property of the specular standard material type in Tonatiuh. In most of the literature I have seen the distribution of the surface errors are defined as some sort of bi-variant distribution using 2 angles to describe the deviation of the reflected ray.

So I am wondering which of the two distributions offered (normal, and pillbox) tend to correspond best with experimental results and how the distributions are defined, (i.e. what does the value given to m_sigmaSlope equate to in terms of how the distribution is defined).

Thanks in advance for your help.

Regards, Daniel

[powelld]

I have noticed that in this document about the raytracing program Soltrace http://www.nrel.gov/docs/fy14osti/59163.pdf This image appears:

I am wondering if it is possible that Tonatiuh describes the distribution of errors in a similar way (whereby the m_sigmaSlope parameter represents the radius of a circle in which the deviation of the actual reflected ray from the specularly reflected ray can be represented).

Am I on the right track here?

[ilescener]

Dear Daniel,

I guess you are referring to the Specular_Standar_Material, which is the model commonly used. In this case, the surface is modelled as it is shown in the first figure. In this model, there is a macro surface normal error, defined by a Gaussian distribution with standard deviation of sigmaSlope. An error of theta in the normal vector (which follows this Gaussian distribution), generates an error of two times this value in the reflected ray. Regarding this, Tonatiuh allows to set up the error in the normal vector through the parameter sigmaSlope. On the other hand, the pillbox distribution has more less the same shape than the Gaussian but the maximum value allowed is this sigmaSlope, you can imagine it as a truncated Gaussian.

Regards Iñigo

[powelld]

Thanks Iñigo!