Rendering Equation

[pat777]

Is it right to say that full GI not only solves the rendering equation, but extends it so lighting is completely realistic? I know algoritms that solve the rendering equation completely are quite close to realistc. However, the rendering equation doesn't contain all light interactions that exist in this world.

 

[bloodbob]

GI is often just refered to for indirect lighting ( inter-reflections ) of course most GI solutions don't have any wave mechanics or quatum mechanics so of course it can't be totally realistic.

So no it doesn't contain all the interactions of the world but its probably makes the biggest differences out of all the changes in the lighting models we have to date.

 

[pat777]

GI is often just refered to for indirect lighting ( inter-reflections ) of course most GI solutions don't have any wave mechanics or quatum mechanics so of course it can't be totally realistic.

So no it doesn't contain all the interactions of the world but its probably makes the biggest differences out of all the changes in the lighting models we have to date.

Thank you, I thought GI calculates both direct lighting and indirect lighting great.

 

[bloodbob]

GI is often just refered to for indirect lighting ( inter-reflections ) of course most GI solutions don't have any wave mechanics or quatum mechanics so of course it can't be totally realistic.

So no it doesn't contain all the interactions of the world but its probably makes the biggest differences out of all the changes in the lighting models we have to date.

Thank you, I thought GI calculates both direct lighting and indirect lighting great.

Lol in involves direct lighting to I didn't bother to add.

 

[GameCat]

If all you need is realistic images, modelling all the interactions of light in the real world isn't necessary. In general there is no need to worry about diffraction for example, or quantum level effects for that matter. A simple geometric model of light is usually sufficient. On the other hand, takoing more wavelength samples than R, G and B might be required to get something as simple as proper transparency.

If you want to read about what is reasonable to include in the rendering equation and what is not, I recommend you read this: edit: wrong url, I'll see if I can find the one I was meaning to put here....
Interesting and well written.

 

[ET]

Damn it, for a moment there I wasn't sure what geometry instancing had to do with lighting. Too many abbreviations!

 

[AlNom]

What does it stand for in this case?

 

[akira888]

Global Illumination

 

[pat777]

If you want to read about what is reasonable to include in the rendering equation and what is not, I recommend you read this: edit: wrong url, I'll see if I can find the one I was meaning to put here....
Interesting and well written.

What about the potential equation?

 

[GameCat]

Sorry, I sort of forgot about this topic. I still haven't found that link... Was a really good one to, a full online book about metropolis light transport and the virtues of path tracing in general. Ah, well.

Anyway, I'm no physicist so I'm not really sure why you'd need to solve the potential equation to get good lighting. Maybe you could educate me a bit on what concrete effects it has on lighting? My point was just that many things that require fairly complex and low level physics models (like fluorescent lamps or glow in the dark paint) can be faked so easily that the burden of doing the extra modelling isn't justified. The tough part is of course to identify what phenomena you can fake and what you can't. Until fairly recently, no one cared about sub surface scattering for example, but now everyone agrees that it increases rendering realism immensly for certain subjects (like human skin, or milk, or marble etc).

 

[pat777]

Sorry, I sort of forgot about this topic. I still haven't found that link... Was a really good one to, a full online book about metropolis light transport and the virtues of path tracing in general. Ah, well.

Anyway, I'm no physicist so I'm not really sure why you'd need to solve the potential equation to get good lighting. Maybe you could educate me a bit on what concrete effects it has on lighting? My point was just that many things that require fairly complex and low level physics models (like fluorescent lamps or glow in the dark paint) can be faked so easily that the burden of doing the extra modelling isn't justified. The tough part is of course to identify what phenomena you can fake and what you can't. Until fairly recently, no one cared about sub surface scattering for example, but now everyone agrees that it increases rendering realism immensly for certain subjects (like human skin, or milk, or marble etc).

Well, I'm a layman on this subject so I'll try my best. I think one of them(rendering equation/potential equation) describes how light emits on to surfaces and the other describes how surfaces absorb light. I'm not sure which is which though. Bidirectional path tracing solves both equations, I believe. The potential equation is adjoint to the rendering equation.

http://www.graphics.cornell.edu/~eric/CGForum.html

As an example of its versatility we derive a Monte Carlo rendering algorithm called bidirectional path tracing that seamlessly integrates the ideas of shooting and gathering power to create photorealistic images.

Edit: I realised you weren't asking about what the potential equation is, but what noticable effect it has on good lighting. I think the potential equation can be an optional alternative to the rendering equation.

 

[davepermen]

Sorry, I sort of forgot about this topic. I still haven't found that link... Was a really good one to, a full online book about metropolis light transport and the virtues of path tracing in general. Ah, well.

http://graphics.stanford.edu/papers/metro/

this one? (hint: google("metropolis light transport").first_hit())

 

[GameCat]

Nope, that's just the paper that introduced it.

 

[pat777]

Nope, that's just the paper that introduced it.

http://graphics.stanford.edu/papers/metro/metro.pdf
That site links to that paper, right?

 

[GameCat]

http://graphics.stanford.edu/papers/metro/metro.pdf
That site links to that paper, right?

Yes.

 

[davepermen]

hm.. was there another one? i thought it was this one i've read (wich was a huge one..)

lets check again.

 

[pat777]

Maybe one day we will have enough computing power and knowledge to compute a Global Illumination algorithm which is 100% physically accurate.

 

[JF_Aidan_Pryde]

Maybe one day we will have enough computing power and knowledge to compute a Global Illumination algorithm which is 100% physically accurate.

Won't that require as much energy as all the matter in the universe combined?

 

[pat777]

Maybe one day we will have enough computing power and knowledge to compute a Global Illumination algorithm which is 100% physically accurate.

Won't that require as much energy as all the matter in the universe combined?

Ok, I just meant a lighting system that's 100% accurate. You don't have to render the entire universe to get a lighting algorithm 100% accurate physically. Stars may be too close and too small when rendered in a scene so it's not physically accurate. However, that's a slight physical problem with the artist, not the algorithm.

 

[davepermen]

http://www.uni-kl.de/AG-Heinrich/Thomas.ig.html

good enough for me.. espencially the night shots..

paper: http://www.uni-kl.de/AG-Heinrich/MediaMLT.pdf