[Radiance-general] daylight factor and rtrace cmd

[Greg Ward]

Hi Thomas,

I agree that the -oo output is odd behavior.  I'm a bit afraid to  
change it, though, because there might be someplace in the code where  
zero ray lengths are problematic, and the initial "false primary" ray  
is needed for consistency for rtrace's reporting routines.  I suppose  
I can try it and see if anything bad happens.

I'm not sure what you mean, though, about the -I option changing the  
effect of -ab.  I suppose it depends on how you think about it.  With  
-ab set to 0, only direct illumination arriving at that position is  
considered.  Setting -ab 1 sends out sample rays over the hemisphere  
to gather indirect contributions.  Is this other than what you would  
expect?

-Greg

> From: Thomas Bleicher <tbleicher at googlemail.com>
> Date: July 16, 2009 10:40:33 PM PDT
>
> On Fri, Jul 17, 2009 at 3:45 AM, Greg Ward<gward at lmi.net> wrote:
>> Yes, the -oo option combined with -I doesn't really work properly  
>> in rtrace,
>> as it "fakes" an intersection with an ideal diffuse surface a tiny  
>> amount
>> (about 1e-6) in front of the point you specified, while setting  
>> the ray
>> origin arbitrarily one unit before it.  No actual intersection  
>> calculation
>> is done, so no worries about intervening rays.
>
> So the 1 unit offset is only a fake to make the output options happy
> and does not interfere with the scene geometry. That's good.
>
>> This is something I should
>> fix, but since you give the origin on input, it hasn't come up  
>> very often
>> where people want the same origin on output.
>
> It's a bit inconsistent to throw a file at rtrace specifying what  
> you want
> and getting back from rtace what it actually did. Like ordering a door
> and getting delivered some planks and a saw. (Should we call this
> 'IKEA rendering'?)
>
> While we're at it, I think that this virtual intersection also  
> swallows one
> bounce out of your '-ab' specs. That only becomes obvious when you
> try to eliminate indirect contributions ('-ab 0' and '-ab 1') but  
> then it's
> rather annoying when you don't remember it.
>
> What's the community's opinion? Should rtrace raise the bounce count
> for '-I' internally or should it just be well document? I'm not  
> sure what's
> more consistent with the way Radiance behaves now or if there are any
> side effects.
>
> Thomas