[Radiance-general] Pixel to solid angle transformation

[Greg Ward]

Yes, although an equisolid-angle fisheye has been requested by users, since it matches commonly found lenses, it is pretty low on my priority list.  All of the existing view types in Radiance vary solid angle somewhat with image position.  The means used to calculate solid angle in pcomb involves taking the area contained by each pixel as its rays are projected onto a unit sphere, thus leveraging the existing image-position-to-ray calculations already present in the code.  These latter routines are rather difficult to get right, which is why adding new view types is a bit painful.  (It took over a year before all the bugs were worked out in the latest -vts addition.)

Best,
-Greg

> From: Marija Velickovic <maricanis at gmail.com>
> Date: February 15, 2013 2:15:37 AM PST
> 
> Hi Lars,
> 
> You are right the math behind is what interests me.
> I know that it depends on projection. For the moment I'm interested in angular fisheye view used for evalglare input images. I thought that is gives equal solid angle for every pixel, but it seems that angles vary slightly (increasing towards image corners), and that's what is puzzling me.
> 
> Marija
> 
> On Thursday, February 14, 2013, Lars O. Grobe wrote:
> Hi,
> 
> I think Marja wanted to know the math behind, not the tools available to query a pixel's solid angle. Unfortunately, the calculation depends on the projection, and there are several projections included in Radiance now. One simplification may be to reproject anything to e.g. an equisolid fisheye, where the solid angles of all pixels should be equal and could be calculated as the solid angle of the image divided by the amount of pixels.
> 
> Cheers, Lars.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.radiance-online.org/pipermail/radiance-general/attachments/20130215/9f172cb2/attachment.html>