Understanding Limit Reflection

Dear Group

I am wondering if anyone could give me an in depth explaination of how the -lr parameter in rpict works? I have noticed that setting high can greatly increase computation time so I wish to find out how to better optimize this parameter. I know there was something in "Rendering with Radiance" on this but unfortunately my book is packed away in a box as I am moving right now. Does this parameter affect the ambient calculation? If so then how? When I read the description in the Radiance manual it simply says "Limit reflections to a maximum of N". Does this limit the amount of ambient bounces set with -ab? May I ask does this parameter deals with the virtual light source calculation only? If I do not have many (if any) virtual light sources is setting this value to 1 or 2 acceptible? Does setting this value this low greatly adverse to the accuracy of the calculation if there are not many virrtual light sources?

Regards

Marcus

···

_________________________________________________________________
Get MSN 8 and enjoy automatic e-mail virus protection. http://join.msn.com/?page=features/virus

Hi Marcus,

-lr limits the reflections form any 'normal' specular surface (e.g. a plastic
with fourth parameter greater than zero). If light from sources hits such a
surface, its reflection doesn't get considered in the direct calculation
anymore (you need the mirror material which produces virtual sources
instead). However, the reflected light gets considered within the ambient
calculation, so the -lr setting has an indirect influence on the accuracy of
the ambient calculation.
Otherwise, -lr is completely independent from the ambient settings and the
virtual source settings, i.e. it doesn't interfere with -ab or -dr (number of
relays from virtual sources)

Often the additional specularity of normal room material surfaces doesn't
contribute that much to the illuminance, with exception perhaps of direct
sunlight shining on a polished tabletop with glossy finish or the like. I
remember this because of my experiments with the direct cache. In such a
special case like above, for example, I observed relative errors of around
10-12 % occuring when ommiting the specular part completely (as if one would
set -lr = zero).So setting -lr = 2 shouldn't do much harm normally (I think
the default is 6 ). A different thing however is the generation of those
typical weird images with opposing specular surfaces where you see the
reflection of the reflection oft the ... etc, in this case -lr needs to be
(almost) infinite, of course. ( I once saw something on a website of one guy
playing around with this feature, I forgot the url, you may ask pillo, I
think he gave me the link..

Another hint: if you set a very high -lr, beware of -lw (weight), too, which
is lurking in the background, often unnoticed...:), and cuts off every ray
which doesn't contribute more radiance than the -lw setting to the
calculation.

-Carsten

Hey, I resemble that remark!! :wink:

Rob Guglielmetti
[email protected]
www.rumblestrip.org

···

On Saturday, August 23, 2003, at 06:04 PM, Carsten Bauer wrote:

Another hint: if you set a very high -lr, beware of -lw (weight), too, which
is lurking in the background, often unnoticed...:),