PDA

View Full Version : Confusion with Caustics...


Integrity
03-18-2005, 12:11 AM
According to the online version of Webster's dictionary...

caustic
Function: noun
Date: 15th century
1 : a caustic agent: as a : a substance that burns or destroys organic tissue by chemical action b : SODIUM HYDROXIDE
2 : the envelope of rays emanating from a point and reflected or refracted by a curved surface

...and Newtek's Lightwave feature description...

Caustics - fast and accurate light reflection and refraction

All in all it obviously explains what it is but what I am downed about is that everywhere I look people only really use this solution to render precise highlight spots...the bright spot created from the reflection/refraction of a surface. Like a glass on a finely stained and protected wood tabletop that gives birth to a bright spot on the table. Note only "bright spot".

I was testing out Lightwave those couple of months ago to see if my curiousity on something was useful. Which leads me to my confusion...correct me if any of this before and after is wrong.

Think of the glass scenario. Then think of how a lens works to project an image onto a sensor. The glass is creating a highlight/bright spot due to the refraction of the light. The lens is doing the same thing.

My confusion is the question is if the image being created from the lens could be called a "caustic"? And if it is, than are caustics really a term for the extreme refraction that creates a focused image on a surface?

Those test with Lightwave were just to see if I could create a lens in Lightwave. Through them I found out Lightwave does not calculate specific things correctly. I set up a model of a lens with a plane on one side and a light projecting an image on the other. As far as I can remember it did not recreate the image on the plane (I didn't expect it to). But when I shot a narrow beam from a spotlight through the lens it did refract the light back onto the plane correctly, which would have led me to creating something like a scanner for the image (CRT).

As far as from what I have learned...everything you see is a reflection...even what you would call a non-reflective object (something that's not a mirror), it is still reflecting only in a blurred way due to the diffuse scattering of the rays. Specular is the opposite of diffuse, "sharp" reflections...the geometry being smooth. So is caustics a term used for an image of light propagating from a surface and not appearing on it?

playmesumch00ns
03-18-2005, 09:45 AM
Essentially you've got it. Caustic relfections/refractions in the general sense for the focused "hotspots" you see bounced onto a diffuse surface when shining a light on a specular reflective or refractive surface.

In rendering terminology this is defined as a LS+DE path. L is the light source, S+ is one or more specular bounces, D is a diffuse bounce and E is the eye.

i.e. a caustic is what happens when a photon leaves a light source, is reflected or refracted by a shiny surface one or more times then bounces off a diffuse surface before ending up in the camera.

I think your hunch about a projector is correct; a film projector operates by the same method: light is emitted by the bulb, is filtered by the film, refracted by the lenses, then bounced off the diffuse cinema screen to get to our eyes.

jeremybirn
03-18-2005, 11:15 AM
My confusion is the question is if the image being created from the lens could be called a "caustic"? And if it is, than are caustics really a term for the extreme refraction that creates a focused image on a surface?

I think you have the main idea.

Caustics refers to indirect light (light that has been transmitted from another object), not direct light from a light source, and not just a raytraced reflection or refraction.

If you think of global illumination in general being light that has already bounced off another object (or been refracted through another object), then caustics are the specific type of global illumination that calculates the light that has been transmitted in a glossy or specular manner, so that it stays fully or partially focused in the way it is cast. If all you wanted was indirect light that was scattered diffusely, then you could use global illumination without caustics.

-jeremy

cpnichols
03-21-2005, 03:04 PM
I think everyone here has offered a pretty good explanation on what could be considered a slightly confusing subject. But let me try to explain it in a different way...

Caustics are part of Global Illumination. Think of its counter part as being Photon mapping. Photon mapping basically calculates how diffuse light bounces, and Caustics calculate how Specular light bounces. More specifically caustics basically calculate how the ray is bent when reflecting or refracting off a glossy surface. Take a typical indoor swimming pool situation. The light source hits the tinny ripples of the pool, and based on the angle of the ripples the light the rays reflecting off the wave at slightly different angles. In order to simulate this on the computer, the rendering engine creates a caustic map. Think of it as an image of pixels all space out evenly, where each pixel is a photon. The raytracer then calculates out that ray bounces off the surface and where it hits the wall. Since you can never have enough photons, all sorts of averaging and filtering is done to the result to make it look as smooth and noise free as possible.

Actually, in an ideal world, all Global Illumination, including diffuse calculations (not just specular) could be done with a caustic calculations. It would just be WAY to expensive to do that since you would need to to shoot nearly infinite rays. That is why photon mapping is simply a more specialized version of a Caustic map which is optimized for diffuse light bounces.

In terms of your confusion, I would say that caustic is the image that is on the wall of the pool (in my example). Meaning it is the specular light that is the result of the raytraced reflection (or refraction).

jeremybirn
03-21-2005, 05:15 PM
Caustics are part of Global Illumination. Think of its counter part as being Photon mapping. Photon mapping basically calculates how diffuse light bounces, and Caustics calculate how Specular light bounces.

Mental Ray and several other renderers use Photon Mapping to compute both caustics and diffuse global illumination.

-jeremy

cpnichols
03-21-2005, 10:34 PM
Mental Ray and several other renderers use Photon Mapping to compute both caustics and diffuse global illumination.

-jeremy

Well true, since both methods start with the same mapping of photons (or rays), one could use the rays for both. However, what the rendering engines does after those rays are calculated is different. Under your example, where photon mapping is used for both caustics and GI, one could say that caustics is the specular part of the photon mapping, and GI is the diffuse. N'est-ce pas?

Ian Jones
03-21-2005, 10:56 PM
I don't see the point of arguing about if caustics photons are the 'specular' photons. This type of rhetoric adds confusion, and specular imo is an old term anyway. Everything is reflection, unfortunately cg has to simulate it somehow and thats where the terminology jungle begins.

jeremybirn
03-22-2005, 05:00 AM
I don't see the point of arguing about if caustics photons are the 'specular' photons. This type of rhetoric adds confusion, and specular imo is an old term anyway. Everything is reflection, unfortunately cg has to simulate it somehow and thats where the terminology jungle begins.

The difference between diffuse, glossy, and specular reflection is one you see in the real world and in computer graphics - the fact that words like caustics and specular have been around for awhile doesn't make them unimportant.

Caustics simulate indirect specular or glossy light reflection. I don't see any arguing about that - everyone who has posted answers above has tried to explain that.

-jeremy

playmesumch00ns
03-22-2005, 08:45 AM
specular imo is an old term anyway. .

No, specular is a very useful term. When light reflects off an object it's either diffuse, or specular or a mixture of the two (glossy). Specular in "proper" terminology means a light bounce directly along the reflection direction.

playmesumch00ns
03-22-2005, 08:47 AM
Caustics simulate indirect specular or glossy light reflection. I don't see any arguing about that - everyone who has posted answers above has tried to explain that

NO! YOU'RE WRONG DAMNIT! Caustics are indirect light reflecting in a glossy or specular fashion!

:)

cpnichols
03-22-2005, 04:00 PM
NO! YOU'RE WRONG DAMNIT! Caustics are indirect light reflecting in a glossy or specular fashion!

:)

Wait isn't that the same thing? ;) BTW, to the original poster, I hope that through all the "noise" we have answered your question in some way.

Integrity
03-22-2005, 06:38 PM
I'd have to say that after looking more into the "terminology" and reading your replies (thank you for all of your help), I think my confusion has disapeared.

It is amazing how formal...but also how rude the dictionary of a language can correct/prove someone's knowledge of a term:

spec·u·lar
Function: adjective
: of, relating to, or having the qualities of a mirror

dif·fuse
Function: adjective
1 : being at once verbose and ill-organized
2 : not concentrated or localized
Function: verb
transitive senses
1 a : to pour out and permit or cause to spread freely b : EXTEND, SCATTER c : to spread thinly or wastefully
2 : to subject to diffusion; especially : to break up and distribute (incident light) by reflection
intransitive senses
1 : to spread out or become transmitted especially by contact
2 : to undergo diffusion

glos·sy
Function: adjective
1 : having a surface luster or brightness
2 : attractive in an artificially opulent, sophisticated, or smoothly captivating manner

caus·tic
Function: noun
1 : a caustic agent: as a : a substance that burns or destroys organic tissue by chemical action
2 : the envelope of rays emanating from a point and reflected or refracted by a curved surface
Function: adjective
1 : capable of destroying or eating away by chemical action
2 : marked by incisive sarcasm
3 : relating to or being the surface or curve of a caustic

Not to distinguish anyone but jeremybirn pretty much answered my question.

But that very knowledge of terms from different people is what confounded me. In sense the dictionary doesn't matter because we are all brought up learning words in different ways. With the added art of CG people change terms to fit certain things or create slang to describe things thus pretty much burning the dictionary, which is where A LOT of the arguing on these kind of forums come from. If I am totally wrong than just say so, but this is what I have seen.

Which is why I'd have to say that you are all correct. What I really wanted to know was the term defined by the people and not by the formal dictionary, so I will know what I am doing when it comes to creating stuff...so I know that I am using the right terms that everyone will agree upon.

Overall...from the thriving community...and in somewhat of a complete aggreement with the dictionary...this is what I have set in my mind and learned...please correct me if I am wrong...but I also ask that you think about the different possibilities of the terms...

Light emanates from an energy source creating photons (in creation not transmission in any way), due to the finite way a processor can calculate something because of our perception of temporal space...we have to crunch down the what seems to be an infinite amount of rays (from what I have learned rays are just a way to visualize things...a path, technically there is no "ray", just a photon being created/shot out from the energy levels in an atom or being transferred through them). When they come in contact with something (ignoring minute scattering from air molecules), they either reflect, go through (I forget the technical term), or are absorbed. When they are reflected, they always follow a law (he he, once again I forget the name), the incident ray is equal to the reflected ray. The microscopic geometry is what gives birth to either diffuse or specular appearance (note that I use the word appearance), or glossy which what I'm going to take from playmesumch00ns is in between the other two. Photons going through the material is transparency along with the density of the material which is governed by the refraction index, and absorbtion which will give color (well along with the reflected ones too, it's all of them that give it's final color). Once the photon goes through all of this...from what I have gotten here...a caustic is when those photons are reflected/refracted back from a curved surface (through dictionary) and are focused back down to create an image on whatever surface they come to next.

From this I am going to consider diffuse, glossy, and specular the adjectives they are to describe the appearance of the reflected rays from a material. Caustic and (wow I can't find a term to decribe the opposite, I'll use diffusion, if there is a term please say so but GI just didn't sound right) diffusion are the appearance of the focused/unfocused images that are reflected back off another material due to indirect light.

This is where the whole things began really. Everywhere I looked on forums and tutorials I saw caustics only in appearance by a "hotspot" or bright patterns like in the pool example that cpnichols gave. It all made sense but I really wanted to clear up the "people's" definition to ask if an image created from a lens for a camera could be called a caustic. It sort of put a final close on my confusion.

Thank you.

cpnichols
03-22-2005, 06:52 PM
Actually what you say makes sense of course in terms of photon scatering etc... I used to think the same thing until I saw a documentary on String Theory and all that went out the window. But that is WAY OT.

playmesumch00ns
03-23-2005, 08:40 AM
Yep think you've basically got it. Thinking of it in terms of photons is helpful since it's the closest thing to how light behaves in the real world (ignoring diffraction etc).

The word for what happens when a photon goes through something is transmission, which can itself be diffuse or specular (e.g. frosted/clear glass).

By "the opposite of caustic", which you call diffusion, I'll assume you mean light that does a diffuse bounce and hits another surface. The term for this I see most frequently is diffuse interreflection, which is nice because it's just adding an extra bit onto "diffuse reflection", which is, after all, what's happening. Diffusion has several different meanings, but in terms of rendering it normally has to do with the scattering and absorption of light in a medium such as the atmosphere, smoke or even skin.

I think the terms we use everyday have been bastardised to cover a whole gamut of different effects. As the technology develops and we increasingly take our cue from the sciences of radiometry etc. the precise definitions of words becomes increasingly important. Science is the only business where techno-jargon is justified :)

Ian Jones
03-23-2005, 01:03 PM
I think the terms we use everyday have been bastardised to cover a whole gamut of different effects. As the technology develops and we increasingly take our cue from the sciences of radiometry etc. the precise definitions of words becomes increasingly important.

This is what I was referring to, the semantics of language. I didn't mean to imply that the term 'specular' is irrelevant. I just re-read what I wrote and I can see I made a mess of what I was actually thinking.

Leigh says in the texturing workshop thread: ' Specularity is basically a way of faking the reflection of light on the objects surface '

'Specular' has also been referred to as a way of describing a surface as 'reflective, with properties like a mirror'

So I can see two ways of referring to caustics:

Caustics = specular GI
Caustics = reflected / refracted GI

IMO, because of the different uses of the word specular in cg terminology the latter is open to less interpretation, and avoids the possible confusions. I hope you can see my point of view, perhaps my understanding is lacking... but I thought I'd put the idea out there for discussion.

cpnichols
03-23-2005, 03:42 PM
I think the terms we use everyday have been bastardised to cover a whole gamut of different effects. As the technology develops and we increasingly take our cue from the sciences of radiometry etc. the precise definitions of words becomes increasingly important. Science is the only business where techno-jargon is justified :)

Yeah but I hear people bastardize things all the time, just to use the the "cool" new technical term. I have even heard supervisors say: "I want to see more fresnel on that object..." when what they really mean is that they want it to be more reflective.

As the rendering software starts to become better and better at recreating real world lighting, it is important that we use real world terminology, as you have pointed out. And while I come from a scientific background, as a lighting artist today, I find it more important to adapt to the same language that they use on set (in the real world). That means using the same language as they use in cinematography.... and the term Specular and Diffuse LIGHT are used all the time... and the term Shinny and Matte SURFACES are used all the time. Sometimes simple words give the best results. They are not as sexy, but there is a reason they have been around for so long. But that is just me, some people like to use more scientific terms such as caustics and fresnel, and those are valid too as long as they are used correctly.

CGTalk Moderation
03-23-2005, 03:42 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.