[Radiance-general] lighting fixtures data

[Greg Ward]

Hi Will,

I think your problem is in "source_pic.cal":

mr(r,g,b) = r;
mg(r,g,b) = g;
mb(r,g,b) = b;

myu = Px/128;
myv = Px/128;

First off, you are using Px as indices for both u and v picture  
coordinates, which is bound to be wrong.  Second, you should probably  
be using something based on the ray direction rather than the  
intersection point, which is with the light source surface and  
unrelated to the output distribution.  Something more along the lines  
of:

myu = atan2(-Dx, Dz)/(20*DEGREE) + 0.5;
myv = atan2(Dy, Dz)/(20*DEGREE) + 0.5;

The above assumes that your light is pointed in the negative Z  
direction (as it seems to be) and that your distribution is oriented  
with the horizontal scanlines along the X-axis in world space.  It  
also assumes that the picture is square and covers 20 degrees  
horizontally and vertically.

A paper that covers this idea in great detail, including local  
photometry modelling, is:

	M. Goesele, X. Granier, W. Heidrich, and H.-P. Seidel
	"Accurate Light Source Acquisition and Rendering"
	SIGGRAPH '03
	<http://www.cs.ubc.ca/~heidrich/Papers/Siggraph.03.pdf>

-Greg

> From: william reynolds <william.reynolds at oriel.ox.ac.uk>
> Date: March 5, 2006 5:54:03 AM PST
>
> hi everyone
> i've been trying out this hdr stuff, and i've run into a bit of a  
> dead end with my understanding!
>
> i've left this on the general list, as it seems a slightly more  
> general use of hdr imaging. correct me if im wrong!
>
> i've tried what you suggested greg, using an image i found (its not  
> hdr, thoug i converted it to .pic) rather than a photo of the real  
> lantern i want to use (hopefully i'll be able to go in and take  
> some photos early in the week). but the images i can generate are  
> turning out a bit messy - i wasn't sure that its just noise because  
> the patteern seems very regular, and there are 2 sorts of "mess" on  
> the image as well. i've put my files up at http://users.ox.ac.uk/ 
> ~orie1226/rad/ if someone wouldn't mind having a quick look at what  
> i've done. (files.txt explains what everything is).
>
> basically i've used the image to define the output distribution of  
> a source, and am then using that to light a room. the likely  
> problems seem to me to be either:
> -some badly set arguments in my .rif file (i dont feel like i  
> completely know what im doing, so this seems quite possible!), or
> -a problem with the quality of the image im using to describe the  
> source distribution. it's not that good quality, but i didn't think  
> it was that bad.
>
> also, all i've been able to find about using hdr images in  
> rendering is about using an image showing the light distbution  
> across a whole scene so that a entire scene can be lit by a single,  
> large sperhical source surrounding it. (at least, i think thats the  
> idea!).
> what im trying to do is define the complete output of source as i'm  
> trying to model the effect of different combinations of sources on  
> my scene. does anyone know of any literature out there that might  
> be relevant to this (i think simpler) approach?
>
> thanks for your help.
> will