You all know that raytracing is an algorythm that uses beams of light to decide the appearence of objects surfaces in a scene. Thus, light and its physical behaviour are the things that let us see what there's in the scene and how it look is.
So, the most important thing when doing 3D graphics is to undestand how light works both in reality and in the simulaiton of the software we are using for rendering our scene. Many hobbist often thinks that the modeling part of their work is the most important, and spent lots of hours tring to add lots of details and precision to their objects.
But the objects in the scene are going to be seen only when light beams hit them, and the look of the objects depends by the characteristics of that light. For instance the type, the colour, the displacements of the lights in the scene will influence deeply the object look, but also the attributes of the surface of the objects that affect the way the light behaves on them.

The modeling of perfect objects is not so fundamental. If the lights in the scene will not be placed the right way to show the perfection of the objects, it all wasted time.

I'm not trying to teach someone how to do raytracing, nor how to assemble scenes. But I'm tring to make people understand the behaviuor of the real light and how it's simulated in raytracing algorythms. This may be interesting (and useful) for someone who has always wondered why some effects exists in real life, or why some cannot be represented in syntetic images.

The first thing to define is the type of the light we want to consider. There are, in fact, quite different kinds of lights, both in reality and in raytracing programs. Lights always comes from a source, be it a lamp a candle or the Sun. Every light source has different characteristics which is then simulated with the software. So we can distinguish between point light, spotlight, extended lights and parallel light sources (TAV 1).

Point light sources do not exists in reality, but are the simplest form of light sources which are heavily used in raytracing both for their semplicity and relative low needs of rendering time. The light sprins from a single point and it's radiated uniformly in a spherical form.

Spotlights are light sources with interesting characteristics as they can be used to represent particular real light sources. The light is contained in a infinite cone, whose vertex is the source. Light then influences only what is "contained" into this cone, while all it's outside is not affected at all. The projected light on a plane is represented by a circle. The intensity of the light into this circle changes uniformly. At the center of the circle the light has its maximum intensity, while going outside toward the edge of the circle it falls off (often following a user specified parameter) till the edge, beyond which the light coming from this source does not exist.

Parallel light beams have not a defined source as they are considerend light beams coming from an infinite distant light source. This kind of light "fills" the space uniformly, and there's often a way to limit the space that contains it in a cylinder (and so it is called cylindrical light source).

It may appear stupid but limiting the influence of the light in a determined region may be useful to speed up rendering time a lot as the number of light beams to be calculated diminishes a lot in the scene.

Extended or area light sources need a separate, deeper analysis as they are the only simulated light source that can describe a real light source. In fact in real nature there are no infinitely small light sources (like point light) but the light comes aways from a surface. This characteristic of real light is very important because it affects the kind of shadows that are cast by the objects in the scene.

Shadows are the socond most important things that affect the final quality of an image. One can often understand that a beautiful image is not real looking at the shadows in the picture. This is a thing that many "artists" are keen to forget.
If you look at the shadows in the real world you can see that there are no shadows with sharp edges. Even shadows cast by the sun are not sharp. Sharp shadows can be obtained in special cases (with special lights), but they are not so common and so they cannot be representative of the real world.
The soft edges of the shadows is due the fact that in the real world the lights sources have always a surface which extends in at least two dimensions (with respect to the observer). So the edge of an object that casts shadow blocks only part of the light beams coming from the light source, while another part of the light beams (always coming for the same light source) are not blocked. This produces the penumbra effect (TAV 2), a thing well diffused in real world that, if not simulated, make the picture looks artificially made (or synthetic).

As the number of light sources increase the passage from dark shadow to penumbra is smoother giving a more realistic effect.

So to have realistic shadows in our scene we need to place area lights instead of simpler (and faster) dimensionless light sources. However, the use of this light may have some problems. The first is surely the fact that not all programs support extended area light sources. This limitation is easily avoided by placing lots of dimensionless light sources one near the other distributing them on a small area. The other problem is surely the raise of the rendering time as more light beams have to be traced, even though the final quality is much greater (and realistic).

The use of the lights in the scene is quite complex, and it would be very difficult to find an equlibrium between regions which are exposed to direct light and regions which are shadowed (that's they do not receive any light), if diffused light were not considered.
As we'll see later, every object reflects in all directions parts of the light it receives. This diffused light then is received by all the other objects around the first one, and in its turn the object receive also the diffused light of all the other objects in the scene other than direct light coming from light sources. The final effect is the lighting of shadowed regions that are now no more so dark. In real life this is what let us look under a table and see the floor even though the light source is above the table itself.
But this infinitely long process, which really happens in real life, cannot be simulated with a computer in a finite amount of time. So all rendering programs have a parameter, called ambient light, which is used to simulate this kind of effect. The ambient light is used to light the dark part of the scene making them not look completely black and hiding all the objects they may contain. This is a quite realistic and expecially fast way to simulate the complex process of diffusing light beams all around, even though it has some limitations we'll se later.

Till now I've described in a easy way how real light is simulated in a rendering program. However we have also to know how light work in real life to understand some of the effects we see everyday and which we are not able to make our program reproduce in a syntetic image.

First of all, what is exactly a light beam? Without going into too complex description, a normal natural light beam (like those coming from the sun) is a "group" of different simple sinusoidal waves with different frequencies (or wavelenghts) and phases. This "group" defined as wave packet is a mix of all the colours (a colour is described by a simple sinusoidal wave of a determined wavelength), and sun light beams contain all the wavelenght of the optical spectrum (see later) so it is able to give colour to all the visible objects we can see. Beside the natural sun light, there are other kinds of light very rare in nature, most of which are produced artificially, like light of a determined single colour, polarized light, and LASERS. However these will not be considered here, expecially because raytracing engines are not able to simulate their behaviour (but in few interesting cases).

Now a brief description of the physical characteristics of light.

You may wonder why going near a light source the intensity of the light raises.
These behaviour is the effect of a physics law that wants the light to dimish its intensity as it travels away from the source. In real life the intensity of the light is inversely proportional to the square of the distance traveled by the light from a considered point. This simply means that an object which is 2 meters from a light source receives a fourth of the light with repect to an object wich is 1 meters away from the light source. This is due the fact that the light beams distribute themselves on a larger (spherical) area as they go away from the source. Thus, the light coming form a light source or diffused by a surface in all directions diminishes its intensity in a rapid way.

This effect of light intensity decreasing with the distance traveled is not always taken into account by the rendering engine, and so the dispacement of the lights in the scene must be studied a lot to try to give the final picture as much realism as possible.

Other characteristics of the light have greater importance over the appearence of surfaces. These include how light is reflected, refracted, diffused and absorbed, which depend on the type of the light and the type of surface.

Optical light, that is the light the human eyes can see, is an electomagnetic radiation whose wavelenght is in a range extending from 4*10^-7 meter to about 7*10^-7 meter (where ^ means power function) of a very big electomagnetic spectrum. Just below this interval there are the so called infrared waves, while just above this interval there are the ultraviolet waves. The colours we can see are due to the fact that light beams may have different wave lenght. I'm not going deeper into this, but you may be interested that the white light is just the sum of all the wavelenghts in the optical interval. When the light hits a coloured opaque (that's not transparent) surface part of the light, well defined in its wavelength, is absorbed by the surface and the other part is reflected. The light absorbed is subtracted by the total hitting light and the resulting reflected light defines the color of the object. For instance a red surface absorb all the wavelengths but the red one, a black surface assorbs all wavelength, while a white surface does not assorb any wavelength but reflect all ones. This also explains why black objects are always warmer than white ones when exposed to the sunlight, as they assorbs all the rays energy directly, while white objects reflect all rays.

The reflection process is quite important as it's what describes the surface of the objects. It's fundamental to know that the light reflects at the same angle it hits the surface (TAV 3). Together with direct reflection (easily visible with very shining surface like mirrors or metals), there is the diffusion. This is a process, visible mainly on rough surface, that scatter part of the incoming light into all directions with different intensity along the mainly reflection angle. As the diffusion is larger and is relevant also at relative big angle with respect to the main reflection angle, the resulting effect is the distribution of uniform light intensity on the area around the ray beam. If, on the other hand, diffusion is contained in small angles along the main reflection beam, the max light intensity is concentrated in a small area producing the so called specular reflection, cause of sharp white (or saturated) reflections on smooth surface called highlights. Diffusion is what make a shining smooth surface a mirror, while a sheet of paper unable to behave like a mirror. More precisely the mirror has a very small diffusion capacity, while the paper has a quite high diffusion capacity, making it unable to reflect directly light beams without scattering it all around.

But as said before, no computer is able to simulate with a physical model based on optical laws the diffused light. Thus the programs simulate the diffuse effect in other much simple ways. As seen before the ambient light is used to distribute light intensity more uniformly in the scene; but there are also the other effects, such as specular light relection, that can be more or less concentered along the main relfection beam direction.
The influence of such behaviour is quite important: as we have seen before it defines what kind of surface we have (rough or smooth), and so if there's the reflection of other object on the surface (mirror) or not (rough surface).
This effect has been divided into two different effects in most rendering programs using two or more parameters. The first effect defines the kind of reflection the surface will have, that is if it will have sharp or blended highlights. The paramenter acting on this effect is usually defined as specular reflection or highlight. The second effect is the actual reflectivity capacity of the surface, indicating how much of around envinroment it will reflect. Some other parameters may be present to finetuning the final aspect of the surface, like brightness or luminosity, but they are just a mean to simulate effects that the rendering engine is not able to produce by following real optical laws (so they shoul be considered a minus than a bonus). The separation of the effects allows for the creation of not natural (or however existing) materials; is then important to try to understand how the effects work alone and together to be able to create surfaces which are as close to the reality as possible (if we really want to simualte a real envinroment). For instance, no real rough surface with an intrinsic high roughness (low reflectivity) can have fine saturated highlights.

Other characteristics of the surface is its transparency, which have quantity and quality parameters. Quantity refers to how much transparent an object is (from complete opaque to completely transparent like glass), while quality indicates what the light has to do when passign through the surface (and in the material beyond it).
Quality includes two different aspects of the transparent object: the index of refraction (called IOR) and how the colour of the light which pass through it is affected. We'll see refraction and it's characteristics (which are very interesting) in the next issue. When light pass through the transparent surface of the object its colour its altered being affected by the so called (text) effect. This process alters the colour of the light by subtracting all the wavelength but those of the surface colour from the passing-through light. However there may be the possibility that the passing-through light components must not be affected by the colour of the object. This is useful for the simulation of some real complex materials.

Tranparency together with colouring of light (and also ambient light) can be used to create special conditions of real life. Much attention has to be payed to the different coluring of the objects when hit by a coloured light. This has to do with previous description of the absorbing of wavelengths. For instance an object which appear as blue under white light, that is it assorbs all wavelength but the blue one, is going to appear black under a red light, which has only red wavelength (no blu wavelength to reflect). If colours involved are not pure (more wavelength make up the colour of the objects and/or of the light source) strange (but predictable) combinations of colors can be obtained.

I'll stop here, hoping this has been interesting for someone. Next time we'll see what is refraction and what are its effects, dispersion, what really are rainbows, TIR (Total Internal Reflection), iridescence and what is a LASER and how to simulate coerent beams of lights, the use of ambient effects (dust, fog, smog) and a description of what are lens flare and why not to use it everywhere.