[Radiance-general] applied material to all faces of a genbox

[Greg Ward]

Hi Steve,

For a sky view, folks usually apply the fisheye.cal mapping using a  
fisheye capture as input.  Mapping appropriate sky distribution  
values is a challenge, although not one you have to worry about if  
you've managed to actually capture a good HDR sky.  (This is a tricky  
proposition -- see <http://gl.ict.usc.edu/skyprobes/> for examples.)

I remember discussing with the Visarc folks ways to use the output of  
gensky or gendaylit to "correct" a sky image that was inserted for  
purely artistic reasons.  Simply using the color from an image and  
the luminance from gensky doesn't work that well, since it would make  
eliminate important luminance changes at cloud boundaries and the  
like, I would think.  However, you should be able to combine the  
highest luminance frequencies from your sky photo along with the  
color to get a reasonable impression.

I took the following image:

	http://www.camerantics.org.uk/wallpapers/clouds.jpg

and ran the following:

	djpeg clouds.jpg | ra_ppm -r > clouds.pic
	pgblur -r 40 clouds.pic | pcomb -e 'hfs=1/li(1);ro=hfs*ri 
(2);go=hfs*gi(2);bo=hfs*bi(2)' - clouds.pic > hfclouds.pic

This should give you a picture that contains all the color and the  
high frequency luminance changes of the original, normalized to an  
average of 1.0.  Multiplying this against a sky distribution will  
give you the appearance of clouds while maintaining the overall  
luminance distribution.  Of course, there are caveats, as the high  
frequencies (and even the color) may affect your rendering due to  
their local influence on the output of gensky or gendaylit.

(Note: the djpeg program is not included with Radiance -- you may  
need to use Photosphere or Photoshop to convert a JPEG image to HDR  
format.)

I hope this helps.
-Greg

> From: steve michel <smichel_designer at hotmail.com>
> Date: May 24, 2008 4:04:25 AM PDT
>
> For any interested , I too found that documentation useful to  
> 'unblock' my initial difficulty with image mapping.
>
> BTW Matiu Carr's Image mapping has a snippet of the picture.cal  
> function code. This documentation is helpful but I found out the  
> latest picture.cal packaged with radiance3.8 file has commented  
> lines in picture.cal describing what each item does in the function.
>
> Now Im wondering if there is a valid way of mapping a panaromic pic  
> for a view through those wonderful daylight views we love so much.
> The doc shows mapping on a tube with cyl.cal as way of doing just  
> that. But that might interfere with skyglow and ruin the lux  
> readings? Would a HDR mage mapped to the sky itself with sky  
> luminance values 'indexed' to the bitmap's rgb values achieve a  
> more accurate result??
>
> Steve
>