Indigo Renderer Forum

The problem relies on the fact that when camera and light are outside a sss material with ior different that the one that is surrounding light and camera, the ray paths do not know were to hit the light and camera, so sss volumetrics can be very slow to converge, especially when the light source is small.

But there is no reason for the paths not knowing were to go if the object has the ior of 1 and the rest of the scene the ior of 1 as well.

So it cannot be difficult to enable this in indigo.

That shouldn't actually be a problem I always thought...

But take care to set the precedence of the material right!
Indigos "Air" has a precedence of 1, others should have more.
The Volume with the higher P value cuts in the one with lower value.
Same values for intersecting volumes do mess with the system

Also, using sss defeats the reasoning according to wich "the ior is 1 therefore the ray knows were to go". Scattering is changing the rays direction at any time and some are even dying within the volume. This makes sss a slow process.

Besides, I'm not sure that Indigo would make assumptions when the medium IOR would be 1.0.

CTZn wrote:Also, using sss defeats the reasoning according to wich "the ior is 1 therefore the ray knows were to go".

I do not agree. Rays do know were to go ->in a straight line ->if the ior is the same on the start and on end of the ray.

CTZn wrote:Scattering is changing the rays direction at any time and some are even dying within the volume. This makes sss a slow process.

Scattering does not change the ray direction, it just split the ray, if the object do not change ior.



It is an ancient issue that I spoked before.
But I have thought about it more and look a paper about unbiased rendering and I reached the conclusion that there is no reason for the fog to work grate inside camera and outside very slow. Am I wrong? If so give me a reason.

On the first post you can see the difference between inside and outside:
http://www.indigorenderer.com/forum/vie ... 11&t=10286

Hi, ior is right,
currently in Indigo in bidir mode, connections are not made through IOR1: IOR 1 interfaces. This can result in noisy renderings in some cases.
I'll try to fix this somewhat soon!

Good point ior, I could have sworn that you were bluffing with me I don't know that much.

OnoSendai wrote:Hi, ior is right,
currently in Indigo in bidir mode, connections are not made through IOR1: IOR 1 interfaces. This can result in noisy renderings in some cases.

Hey Ono.
The noisy renders exist always with this current mode.
If you have large illumination areas the render will not be so noisy.
But if your scene has only a small illumination point, noise will appear depending in how small is the illumination.
Small=noisy.

It is like having some caustics or fog behind a glass, or reflected in a mirror.

In this post it seems that you have implemented what I am talking about:
http://www.indigorenderer.com/forum/vie ... &start=164

When can we see that improvement?

I realized now that since, don't know when, single CPU PT renders fog correctly outside and inside camera!
Here an example: Now we only need that in GPU and BIDIR.

When?

Maybe only in indigo 4 we are getting this feature.

Volumetric fog on bidir and MLT, When?

Soon after 3.6.

Cool

This is in trunk (bidir works properly with same IOR interfaces now), will be available next beta.

Is it to say that bidir will become an option for clouds and atmosphere rendering ? I've got troubles realising what this means, sounds like a must though.

CTZn wrote:Is it to say that bidir will become an option for clouds and atmosphere rendering ? I've got troubles realising what this means, sounds like a must though.

yes, should be.