# Three-Phase Method _ Sky Vector - Daylight Matrix

Urtza,

*Regarding question 1.*

The CIE Overcast sky is simetric and independent of the date and location.
That means that it does not matter if you rotate your model and the sky or
change the location and the date, it will always return the same vector
when fed into genskyvec.

Genskyvec approximate the sky as "finite elements" of constant Radiance. I
understand this is done using Ray-tracing. That means that each patch
(element of the sky vector) will have a different value that correspond to
this approximation.

Finally, the sky vector does not depend on the model just as the sun's
position does not depend on the orientation of your apartment. However, I
understand it is possible to calculate a rotated sky vector. This is faster
than performing all calculations again with a rotated model (i.e. if you
rotate the sky clockwise, it is the same as rotating the model counter
clock wise).

*Regarding question 2.*
Klems divition is always the same divition. However, the direction used to
trace rays (for sampling and calculating the View matrix) will depend on
the orientation of the window. This is the same as saying that the same
camera will give you different images if pointing on different directions.

Best,

Germán

···

2015-07-03 6:44 GMT-03:00 <[email protected]>:

Dear experts,

I am trying to learn three-Phase Method for Simulating Complex
Fenestration.

I have two questions of Sky Vector and Klems divisions of Daylight Matrix:

1) Sky Vector:
I am trying with Overcast Sky. I have done as following for Windows (I
have Ubuntu 14 now, but, I am checking the first steps done by Windows):

gensky 12 21 15 -c -B 0.55866 | perl c:/radiance/bin/genskyvec.pl -m 4 -c
1 1 1 > review/skies/12_21_15.skv

(I do not know if is good) So, I am sure that for each time during the
year it generates the same Sky Vector. However, I think that the
"genskyvec" calculation divided the sky in different parts in contrast with
"gensky" command that generate void brightfunc material. Therefore, the
skyvector does launch different values for each sky division? If I rotated
my model I get different illuminance date for only Sky Vector variable?

2) Klems divisions of Daylight Matrix:
When there is a definition of "kbin(Nx,Ny,Nz,Ux,Uy,Uz)", is it because for
each orientation there are some different values? I suppose that for each
orientation it uses the same divisions.

Excuse me, but, I am trying with a rotated model and I want to ensure that
I understand well the method. Sorry for the inconvenience. Thank you in