I thought FG in LW didn't generate caustics!?

[lardbros]

Hiya,

Been messing with GI stuff a lot today and was running through Exception's Wiki stuff to get more in depth, but have found that I'm getting caustic patterns when using Dielectric materials close to walls.

Is this something that Exception simply hasn't updated in his docs... or am i being stoopid?!?


Oh, and while I'm at it, has anyone got an idea of what the volumetric radiosity and the directional rays do within the Radiosity panels?! (Sorry, posted before i read the updated manual, so will go off and check right now)

[lardbros]

Okay, volume radiosity is simply for volumetrics! But nothing on Direcrional Rays in the new docs!

Wish Newtek would just go full power on the LightWiki rather than bother with their PDF docs. They always look unprofessional and a bit pants. Never enough diagrams explaining things!

[MooseDog]

afaik, the directional rays allows sun-rays to go through window glass panes. if un-checked, your "sun" light will not be considered as contributing to the radiosity solution within the room (as an example). i suppose this applies to all "glass" surfaces.

what i'm unsure about is if this is just a f.g. thing, or applies to all radiosity types.

agreed and another vote for a wiki solution to documentation.:thumbsup:

[lardbros]

But exception says, in his tut, that FG doesn't generate caustics, but i for sure am getting them with my renders! Don't get all this, why does FG without Interpolated on still have all the options available, and then when switched on nothing appears to have changed, but the render results are completely different?!?

I realise FG is an interpolated form of rendering GI anyway, so why is there a choice of interpolated and un-interpolated??

Someone please enlighten me (preferably with nice GI enlightening)

[Matt]

I've never managed to get caustics under control, half the time can't even make them appear! Seems to be a black art!

[lardbros]

These aren't genuine lightwave caustics, just the refraction of the light through a dielectric material. (Guess they're still caustics, but not the type that you actually switch on.) It also happens when you have a lens shaped object, but in exception's example, the light doesn't refract through the lens using FG, only does it using Monte Carlo.

[Ztreem]

FG has always done cuastics, but if you uncheck the new option 'directional rays' you wount get any. With interpolated the caustics will also be interpolated so most of the detail will disapear and it will look blobby and ugly. Most of LW caustic's look kind of crappy by the way. With very long render times you can get decent caustics.

[Andyjaggy]

I've never managed to get caustics under control, half the time can't even make them appear! Seems to be a black art!

amen to that.

[Matt]

It seems you need to have things setup a certain way, at the right angle etc. and I don't know what these things are!

[vfxwizard]

FG doesn't generate caustics, but i for sure am getting them with my renders!

If Directional Rays is set to on FG does generate caustics -sort of.

FG is not capable (by algorithm, this is not a shortcoming in LW) to deal with directional light effects and that's why in 9.2 directional rays were disabled by default for FG. There were some side effects like darker renderings (no specular contribution to indirect lighting) and light not passing through transparent surfaces, so FG regained the ability to trace directional rays as an option.

Think how specularity and reflections change as the observer moves around. In GI, every surface point is an observer. Monte Carlo properly computes all the "viewpoints" for directional rays, and is slow. Final Gather by design ignores directional rays and is faster. If forced to do so FG will compute directional rays as well but light will not bounce as expected.

However Directional rays On in FG are useful to quickly check overall lighting levels, before a final render in Monte Carlo. And for refractive caustics sometimes the result can be good enough to fake the eye.

I have made a quick chart (attached image) that compares FG and Monte Carlo showing how different reflective caustics will look.


I realise FG is an interpolated form of rendering GI anyway, so why is there a choice of interpolated and un-interpolated??

The choice between interpolated or non-interpolated depends on scene content and the look you are after.

Non-interpolated Final Gathering mode in LW is the same as an 1 bounce Monte Carlo. But the indirect lighting for bounces after the first is interpolated from the "preliminary solution" and this is a huge time saving with little or no loss of quality (as long as directional rays are off).

In interpolated mode, you can't really distinguish between the preprocess and the rendering, LW does its irradiance caching interpolation both for secondary bounces and for pixels actually displayed.

In short, non-interpolated FG gives you noise and sharp contact shadows much faster than non-int Monte Carlo if you have many bounces.

Interpolated FG, well, is faster than non-interpolated and lighting-wise produces the same results.

Just for fun, try out this: interpolated or not, Final Gather and Monte Carlo will produce identical (non-interpolated) or near identical (interpolated) results where there is only 1 bounce and the scene contains only diffuse surfaces. However in this particular case, Monte Carlo will be way faster because FG is slowed by the preprocess and cache overhead.

[Exception]

You'd do good reading the radiosity tutorial on my site rather than the lightwiki, as it's a newer version:

http://www.except.nl/overig/radiositytute/radiositytute.htm

It explains directional rays.
When you haev directional rays on, you WILL get caustics from FG, BUT they will be incorrect in terms of color and direction. Now, when you turn direct off, no more caustics from FG.

[lardbros]

You'd do good reading the radiosity tutorial on my site rather than the lightwiki, as it's a newer version:

http://www.except.nl/overig/radiositytute/radiositytute.htm

It explains directional rays.
When you haev directional rays on, you WILL get caustics from FG, BUT they will be incorrect in terms of color and direction. Now, when you turn direct off, no more caustics from FG.

Ahhh, i just assumed they were pretty much one and the same. I'll make sure i check back on your site when any changes to LW's Radiosity changes!

Thanks