Subsurface Scattering

[Roy Hirshkowitz]

Coming soon-- path tracer only.  Architectural applications are probably limited-- but it's interesting new territory anyway.

 

[Roy Hirshkowitz]

Here's a blend of SSS and procedural marble:

 

[akinlolu olugboji]

Will do well in certain ceramic tiles

 

[pun1407]

There is very well function of nxt, thank you! :-)

[Roy Hirshkowitz]

Here's a different use of a similar technology:

[Kevin Lockwood]

Is that fog I see?

[Roy Hirshkowitz]

In this case, confined to a box 1 meter high.  Uses subsurface scattering techniques...

[Aleksander Nowacki]

What's the point of this technology?

[Roy Hirshkowitz]

I'm not sure I understand the question.  Do you mean "what are the applications?"

[Aleksander Nowacki]

I just want to know what exactly your message means - Subsurface scattering - is it some new funcionality in nXt? What's so special on these pictures? What gives this technology?

[Roy Hirshkowitz]

There's an ok summary of subsurface scattering on Wikipedia.  It's an industry standard term so there's quite a bit about it out there.  Here's another decent article.

 

Generally speaking, the lighting in nXt and all previous versions of AR occurs at the surface only.  Subsurface scattering is essentially a volumetric technique, where interactions also occur within the medium.  It's required to accurately model materials which we think of as "translucent".  It's also helpful with materials where the light penetrates the surface a little bit-- such as the marble dragons-- and then scatters on its way out.  The feel of the images above would be very difficult to achieve without this technology.

 

A similar technique can also be used to model other participating media, such as the ground fog above.  This has a very different feel than previous fog implementations.  The technique can be extended to model non-homogenous participating media-- things like smoke or non-uniform fog (clouds?) etc..

 

Downside is that it can take a long time to resolve these materials-- accurate simulations almost require a path tracer approach.

[Aleksander Nowacki]

Thanks a lot for this explanation. Looks promissing.

BTW. I found the Path Tracer the most interesting and valuable feature of nXt. It gives realy soft light (for the first time in whole AR history). But it's not perfect - it's slow - sometimes even hours and houndreds passes don't fix the noisy areas. I'd like to ask you - are you planning to develope this engine? I think it realy worth it.

[Roy Hirshkowitz]

It is being developed-- subsurface scattering is an example of this.  I'm working on a stand-alone version of the engine as well.

If you mean-- will the performance improve substantially?-- I don't think so.  I don't have any magic wands to wave at the moment.  One of the things I'm going for with this engine is simplicity.  None of the really technical items matter to the path tracer-- for example: Glossy switches, daylight portals, ambient lighting settings, etc.. have no effect.  The combination of this simplicity and the calculation of more light paths slow the thing down-- quite substantially in the case of interior daylighting.  The above mentioned stand-alone has almost nothing to set.

I'm also investigating another engine that has the potential for even more accurate simulations-- probably a little slower, though.

On the bright side-- computers continue to get faster.