What happened when using rpict to calculate the irradiance

The manpage of the rpict for -i option writes that:

Boolean switch to compute irradiance rather than radiance values. This only affects the finalresult, substituting a Lambertian surface and multiplying the radiance by pi. Glass and other
transparent surfaces are ignored during this stage. Light sources still appear with their original
radiance values, though the −dv option (above) may be used to override this.

However, I did not quite understand it regarding the sentence “This only affects the final result, substituting a Lambertian surface and multiplying the radiance by pi”. For example, for direct sunlight calculation in sun-coefficient method, the reflectance of indoor surface is modeled as zero, if the irradiance were obtained via multiplying the radiance by pi, then the irradiance should be pi*0. Obviously, the calculated irradiance using rpict is not zero otherwise the material map doesn’t work. I guess whether the indoor surface was replace by a lambert surface of a certian reflectance and then multiplied the radiance by pi.

As the quote says, a Lambertian (purely diffuse) surface replaces whatever actual material was found, and the reflectance of this new material is set to pi to compute irradiance rather than radiance. Does this help?

Thank you, Greg! Your interpretation helps me understand why the irradiance calculation is so slow in sun-coefficient method. However, it also brings me into another question that is if a specular blind were involved, would the specular blind be replaced by a Lambertian surface? My guess is no but I still need to confirm it.

Any opaque (non-glass, non-trans, non-aBSDF) surface gets replaced by a Lambertian material when seen directly (or through a transparent surface) by the virtual camera. Indirect contributions, including reflections off of specular and other materials in the room, will still behave as normal during the ray tree calculation.

Thank you, Greg! I got it.