I have been calculating the average of “Sky Exposure” from a grid of points in an urban outdoor area, using the LB component (HumanToSky). I have noticed that there are other components that also calculate sky percentages under the name “Sky View” like the LB (SkyMask) and (ViewPercent) components.
I understand the difference between “Sky Exposure” and “Sky View”, is that the later gives weighted percentages based on the angle of the sky patch with the point of calculation, but “Sky Exposure” treats all parts of the sky dome equally.
My questions are:
If I want to calculate the amount of sky visible from a point in the outdoors, for the sake of measuring the ‘openness’ of a space, does this mean the ‘Sky Exposure’ is the right choice?
What is the accuracy of the ‘Sky Exposure’ in the (HumanToSky) LB component? As the Sky View in the (Sky Mask) component states that for a density of 3, the percentage of error from true value is less than 1%.
It depends on what the thing is that you are measuring openness for. Are you talking about how open a surface is to the sky (eg. how much of the sky does a window on a building see or how well does a surface radiate heat to the sky after sunset)? If so, then “Sky View” is what you want. If you’re talking about how much of the sky a human experiences while standing in the outdoors, “Sky Exposure” is the better metric.
Assuming you are asking about how many vectors are being used to compute the fraction of the sky that is visible, you can see here that the “Human to Sky” component uses Tregenza vectors:
So it uses 145 vectors per point, which are not subdivided like the sky mask, which subdivides each patch 3 times. Granted, using undivided Tregenza vectors likely also means that your error is less than 1% for any reasonable geometry but it can be greater if you are trying to resolve some detailed geometry like tree branches, which does not align well with the tregenza patches.
I now have another question though, is there a fairly accurate method on LBT to calculate sky exposure for intricate geometry? @chris
I am dealing with different forms of shades that definitely wouldn’t align well with the tregenza patches.
I am thinking something like the ‘fisheye’ image analysis rather than vectors.
If you want to evaluate sky view or sky exposure in a way that you can adjust the number of rays used, you can use the LB View Percent and adjust the _resolution_ input.
Fisheye images from Radiance use ray tracing just like the View Percent component. It’s just that each ray corresponds to a pixel. So just use a number of rays that correspond to the number of pixels you would have had in your fisheye image and the result will be just as resolved as a Radiance render.