The synthesis of image 3D often shortened 3D (3D for three Dimensions: X, there, Z the three axes which form the Repère orthonormé of the Solid geometry) is a whole of techniques in particular resulting from the CAD which allows the perspective representation of objects in on a Moniteur of computer. It currently is very much used in numerical Art in the industry of film, is initiated by the studios Pixar and, since 1994, in much of Video games.

History

The denomination images 3D is the new name given so that one called drawing or perspective painting in since the Renaissance.
Voici an example of perspective sight carried out by Piero della Francesca in 1475. Certain painters help themselves of the Perspectographe, first drafting-machine in 3D , for their pictorial compositions. The difference is that, before the appearance of the computers, the prospect was obtained by graphic methods directly derived from the projective Géométrie. Currently, the drawings are calculated numerically on the basis of the three-dimensional numerical data. This makes it possible to easily change the center of projection and its various parameters. In particular, one can calculate series of projections by moving the point of projection and thus carry out Animation S.

The current data-processing synthesized images rest on the same principles of projection on a plan and are indissociable history of data processing. The synthesized images began at the beginning from the Années 1950, in the United States, and were reserved for research, in particular university. One built a system made up of a cathode tube and a light pen, according to an idea of Ivan Sutherland, Massachusetts Institute off Technology (MIT), for the air control of the air force, then in 1961 one added on the screen a cross to indicate the position of the light pen. One practiced image 2D then, then of the image 3D, more expensive in computing times and financially.

Then the universities also made use of the images 3D, and in 1967, the university of Utah specialized in this field, in particular the professors David C. Evans and Ivan Sutherland, who tried to model various objects such as the car of Ivan Sutherland, and which will found in 1968 the company Evans & Sutherland. Then in 1970, Xerox created the PARK (Palo Alto Research Center), which will work very freely, because without commercial objectives; it will emanate from it from many discoveries that Xerox will not be able to exploit. In 1975 was created one of the most famous images of the computer graphics, the teapot, was become since a traditional object of test for the applications 3D. The teapot which was used as model rests now in Boston Computer Museum, close to a computer which reproduces its image in three dimensions. < Br > Until the years 1980, few people approached this field because of the costs of the material. But the appearance of the personal computers, the such Xerox Star, IBM-PC in 1981, and the Apple Macintosh in 1984, democratized the use of the 3D for the study, the production… The evolution of technology leads to superb simulations of shuttles or rockets by NASA, or of landscapes and faces.

But it is starting from the Années 1990 that the synthesized image and the 3D democratized and developed in an important way, with in particular the arrival of more powerful materials allowing the real-time, like charts 3D at Silicon Graphics on Work stations top-of-the-range then later in the computers general public PC with charts 3D like the 3DFX or in the game consoles like the PlayStation, the Dreamcast. Since, acceleration 3D is integral part of the computer materials.

General information

The synthesis of image 3D calls upon a vector Space. This space is broken up into 3 dimensions on the Cartesian axes, usually named X , Y and Z .

Let us take, in a part, a point of reference, and define directions left-right-hand side (X) , before-back (Y) and haut-bas (Z) . To go from the origin to a given point, it is necessary to make:

• X meters towards the line;
• there meters forwards;
• Z meters upwards.

If one can permute the order in which one can make displacement, this combination ( X , there , Z ) is single: a point of the part is represented by a single triplet, and a triplet represents only one point of the part (see Repérage in the plan and space , Cartesian Frame of reference and analytical Geometry ). This triplet of values is called the coordinates of the point.

The point of reference is called “the origin” and notes usually O , it corresponds to the triplet (0, 0,0).

To move of a value has on the left is equivalent moving of a value − has on the right. To move back of a value B is equivalent advancing − B . To go down from a value C is equivalent assembling − C . (See the article relative Integer ).

Now let us take a simple Volume, a Polyèdre. This polyhedron can be defined by the coordinates of its tops. By the data of a series of values ('' X '' ₁, '' there '' ₁, '' Z '' ₁); ('' X '' ₂, '' there '' ₂, '' Z '' ₂); …; ('' x_n '', '' y_n '', '' z_n ''), this volume is defined.

Let us take for example the eight points

• (- 4, -4,-4)
• (- 4, -4,4)
• (- 4, 4,-4)
• (- 4, 4,4)
• (4, -4,-4)
• (4, -4,4)
• (4, 4,-4)
• (4, 4,4)

these eight points define the tops of a Cube of which stops it has a length 8, and whose center is in O .

One thus represented a cube by a whole of values. This space is also called a three-dimensional or “virtual” matrix in the world of the imagery 3D.

Techniques

The synthesis of image 3D breaks up primarily into 2 stages:

• To model what one wants to visualize or represent. This stage is called modeling
• Effectuer the visualization of what one modelled. This stage is called Rendu

In each one of these stages, a great number of techniques exist. Although theoretically the techniques of modeling and of returned are independent, it goes without saying returned must be able to exploit the modelled data and very often a technique of modeling is closely associated with a technique of returned.

Modeling

Modeling will consist in making store by the computer a whole of geometrical data and characteristics graphic making it possible to represent the model then. This model is usually called Scène 3D, according to English " scene" who means " vue" , " will thus be said; sight 3D". See the article dedicated for more information:

See also: Modeling 3D

Types of modeling

• CSG (Constructive Solid Geometry): the objects are defined by operations (unions, intersections…) on geometrical forms in space (polyhedral, spheres, cones, curved of Bézier, splines, NURBS…). This method is well adapted for creation ex nihilo , since it corresponds well to the way in which the human spirit represents the shape of the objects.

• BRep (Boundary Representation, representation by border): the surface of each object is limited by geometrical forms 2D (generally of the triangles). It is the natural format of exit of the scanners 3D which measure the dimensions of points on the surface of the objects. It is also the representation generally used by the display devices accelerated for the 3D.
• Spline S, where surfaces are described by curves forming a lattice.
• implicit Surfaces, where the objects to be represented are described by surfaces uniting of volumes, generally spheres.
• Voxel S, taking again in 3D the analogy of the digital images where space is divided into regular intervals.

Software of modeling

The process of modeling can be is automatic as for a Scanner 3D, where a program will create a data-processing representation starting from a real object of the world, that is to say manual using a software of edition 3D. In this last case, a Infographiste is responsible for the creation of the model. This software is called according to English to model modellers .

See also: Software of modeling 3D

Among the most known software and most widespread are 3D Studio max, Maya, XSI, and the software open source, Blender.

Returned

Returned breaks up in its turn into several phases.

• Calculation of lighting (one speaks about illumination)
• Projection in the plan of observation.
• Drawing strictly speaking with possible application of textures

It happens that some of these phases are carried out at the same time (in the charts 3D for example)

Engine of Returned 3D

• returned is a phase which consists in transforming space 3D into an image 2D. Returned is generated by un/des program (S) which one calls 3D returned Moteur integrated (S) or not into the software of modeling.

Illumination

• local Illumination

• the algorithm of Gouraud, is a method of returned very simplified since it consists in interpolating on the image in 2D, line by line on the level of the pixels, the values of intensity of the interior of a Polyèdre according to the intensities of its neighbors: it is a smoothing 2D.
• the algorithm of Phong is a shade 3D which makes it possible to have a volume and an effect of matter, according to the source of light and of property of reflection of the light of material in each point of the Polyèdre, by interpolation of the normals at the tops.
• total Illumination
• the Radiosité calculates the luminous energy exchanges between elements of surface of the scene by taking account of their properties of reflection and absorption. That requires to break up the scene into finite elements of surface. That allows naturally soft lightings. This technique associated with the throw with rays and other neighbors allowing the inter-reflections is a technique of total Illumination.

Projection

The objects thus defined by numbers can then be drawn on a screen; the triplets of values ( X , there , Z ) are transformed into points of the drawing. This returned the perspective concept of uses:

• conical Prospect, with break points: the more the object is far, the more it is drawn small; this process is sometimes qualified wrongly “true 3D” or of “natural vision”;
• axonometric Prospect, for of which the riding Prospect and the orthogonal projections (in particular the isometric Prospect, to also see descriptive Geometry ): the size of the object does not vary with the distance; the distance is illustrated by a displacement in the plan of the figure.

Other algorithms of returned are used, holding account in particular owing to the fact that the details and contrasts grow blurred with the distance (atmospheric Perspective).

The techniques of the synthesis of images 3D initially distinguished the algorithms from calculation of the hidden sides which worked in space 3D of the scene and those of the returned photorealist which worked in space 2D of the image (pixels). The algorithms of made current fulfill the two functions simultaneously.

Drawing

With these types of prospect, one associates a type of drawing

• Z-buffer (plug of depth of the pixels), using the algorithm known as of the painter (who paints the scene on the basis of the bottom then successively the increasingly brought closer subjects) to post the scenes made up of polygons or more advanced structures like the Nurb S , and which correctly manages the posting of polygons intermingled with the precision with posting because the layout makes pixel by pixel. It is a method 3D of calculation of hidden sides on a pixel scale because it preserves the depth of each pixel.
• the To launch ray (raytracing, from which the Raycasting is derived), simulating the opposite course of the light of the scene towards the eye. This technique makes it possible to relatively easily simulate the physical phenomena which are reflection and refraction as well as the shadows but creates too pure and unrealistic scenes. In this case, projection is carried out at the same time as the drawing because to trace a ray implies to project it in the scene 3D.
• Others…

Techniques complementary to application of texture ( texture mapping ) are used for returned more realistic effects without to make more complex the models 3D. For example:

• the application of textures photorealists resulting from the real object,
• the Plating of relief ( bump mapping ) which is a local disturbance of the normals on the surface and makes it possible to return the aspect of granulous surfaces for example.
• textures of lighting

With these techniques the problems of filtering appear, which is necessary to eliminate the artefacts.

Alternatives

• the representation by edges, equivalent of the drawing the line, represents only the edges of the objects of the scene requiring only one algorithm of hidden sides. This method of calculating in space 3D is still used in many Modeleur S to have a fast posting.

Basic difference with the 2D

With final, there is an image 2D. This image could very well have been drawn directly in 2D, but it was calculated, generated starting from the model 3D.

The world of the 3D in synthesized images is, indeed, mathematically reality. But the mechanism here is reversed, a such draftsman reproducing on paper paperboard a work carved from every angle. The image in two dimensions resulting from returned from a three-dimensional scene is not other than the by-product of this “telegraphic” virtual world. The possibilities of this technology are infinite, just like an artist draftsman could reproduce reality under an infinity of angles and parameters different.

Contrary to an image traditional 3D, an image 3D volumetric (or voluminal), a value gives to all the points of space (matrix algebras). Those all are connected (or not) between them in order to constitute Polygone S which one commonly calls of the facets. These facets inter-connected between them in their turn, constitute with final the object 3D in question. In addition, other points can be them also inside even object, these even to which one can allot additional specific functions, conferring on the object an impression of mass in addition to volume (e.g.: Full object/hollow Object).

Returned relief

The vision in relief comes owing to the fact that, the two eyes being shifted, an object is not seen at the same place by the two eyes (difference of Parallaxe). The computer can generate two different images, one being seen by the left eye, the other by the right eye, and thus give an impression of relief.

Simple means of creating this “artificial relief” consist in generating only one image, known as Anaglyphe, but containing information for the two eyes, two colors, in general green and red. The spectator has glasses having a green filter on a side (this eye sees only red information, which appears in black, seeing subtractive Synthèse ), and a red filter on other side (this eye sees only green information, which also appear in black).

One can also work with only one initial image, with a distribution of the judicious colors. The glasses described above exaggerate the visual effect on these special images, putting at the foreground the subject of the image in green and the background of the decoration in red. One obtains a stereoscopic illusion of depth thus, as well as relief.

See also: Stéréoscopie

Virtual holography

Nothing prohibits with the current processors to calculate in a few minutes, even in a few seconds, of the simple holograms impressed on films with good resolution. The process exists in fact since the beginning of the Années 1970 (the IBM Systems Journal had provided a specimen of such a hologram calculated in one of its numbers of this decade). The process did not have following the time for reasons of cost, but could re-appear today.

Temporal constraints

Although technology is similar in both cases (Cinéma and Video game), there exist notable differences. The constraint of real-time inherent in the video game does not exist obviously with the cinema. This is why the synthesized images used with the cinema, more detailed, are precalculated one by one, which allows the much better made visual. It is thus advisable to make well the difference between the 3D in real-time and the 3D precalculated. The cinema and the video game are not the only contexts of use of the synthesis of image 3D. Other contexts, in particular that of the numerical Maquette of aspect use hybrid solutions, compromises between 3D real-time and 3D precalculated, which make it possible to obtain in certain cases one made visual near to the synthesis of image 3D precalculated while profiting from the advantages of the 3D real-time.

3D real-time

The 3D real-time is actually used in the video games, but also has many other applications: architectural visualization, medical visualization, various simulations, savers of screens… The challenge technical inherent in this kind of 3D is to obtain the best quality of possible image while preserving a fluid animation, which requires to optimize calculations of posting as well as possible. At the beginning, all calculations returned to the main frame (CPU) of the computers, but the power increasingly larger required to improve quality of the images pushed manufacturers to market charts NCV (Peripheral Component Interconnect) specialized in the 3D. The first available one on PC (Personal Computer) was legendary Voodoo of the company 3DFX. Today, the graphics cards integrate for the very large majority of the functions of acceleration 3D. The use of processors dedicated to the calculation of the 3D created the need for defining API (Application programming interface the) standard ones, making it possible the developers easily to reach the accelerated functions and to the manufacturers to include them in the processors. Both API 3D most widespread are OpenGL and Direct3D (component DirectX), more recent competitor developed by Microsoft.

3D precalculated

The 3D precalculated is used for the creation of special effects and cartoon films. Its goal is especially to obtain a very great quality of image. A concern for the computing times exists, but scale completely different from that of the 3D real-time. Indeed, in the latter, a fluid animation requires that it be possible to calculate more than twenty images a second, whereas in 3D precalculated, the calculation of an image can take hours, even days. Once all the calculated images, they are projected at the desired frequency (24 images/dry for film 35 mm e.g.). They are often generated on what one calls calculation a Ferme. It is an important network of computers connected the ones to the others to multiply the computing power. As each second of animation requires nearly 30 images, that means that for 15 minutes of animations the network will have to generate more than 27.000 images.

The Software S calculating these images, the “engines of returned” ( renderers in English) are numerous. With the passing of years, the software general public improved considerably to refine the grain and also to transmit the real effects on the environment imagined by the teams of originators. On the other hand, of the specialist companies with artists with pointed knowledge in modeling, in lighting, and animation are always necessary. The other most known software is RenderMan, Mental ray, FinalRender, Brazil r/s, V-Ray.

A popular technique of calculation 3D is the “throw of ray” (raytracing). In this technique, a luminous ray “leaves” each point of the screen, and “meets” and “reflects itself” on the objects dispersed in the scene 3D.

The studio of animation Pixar is dedicated only to the creation of films in synthesized images since the Années 1980 and is thus recognized like precursor on the matter. It develops in particular RenderMan, an engine of returned considered as one of most powerful, serving with returned today as all the productions of the studio like with that of a vast majority of the synthesized images of feature-length films (see films in synthesized images and Synthèse of image).

The industry of architecture and interior installation uses more and more these methods to facilitate the sale of future projects. The synthesized images make it possible to transmit a new quantity of data twenty more years ago. One of the most famous architects, Frank Gehry, Canadian based in Los Angeles, uses this software to facilitate the design of the various modules which will be integrated into its very organic buildings.

hybrid 3D

The hybrid 3D aims at improving photo-realism of the 3D real-time by precalculating (and thus while solidifying) certain parameters whose calculation is expensive but whose quality is essential to the realism of the image.

Lighting belongs to these expensive parameters. In 3D real-time, it is sacrificed: not or few shadows, not total illumination. However, a good lighting brings enormously to the realism of the image and it is often what makes the great difference between the 3D real-time and the 3D precalculated. To solidify this parameter can be a problem, it cannot more be changed into real-time, lighting cannot be completely dynamic more. But there are many contexts of use of the 3D real-time in which it is not essential to have a completely dynamic lighting and or one can thus allow oneself to solidify this lighting.

Textures of lighting (Lightmap) are then calculated which encodent lighting. This calculation relatively long (is precalculated, it is not made in real-time) but, once carried out, the information of lighting is available instantaneously. The quality of returned can then approach what one obtains in 3D precalculated while bringing interactivity in the adjustments of all the other parameters.

Living rooms

• the Siggraph is the annual appointment of the research laboratories on the synthesis of image.
• Each year, the festival of the synthesized images Imagina takes place with Monaco.

Schools

• In Belgium, the schools of 3D (or computer graphics) most known are the High School Albert Jacquard de Namur and the INPRES of Seraing.
• In France, one can quote, inter alia, Supinfocom, the Goblins, Explosives magazine, the ISAA…

See too

• 2D

• To launch ray
• 3D real-time
• Software of modeling 3D
• Driving 3D
• Stéréoscopie

External bond

Random links:

1949 | Krak | Chassors | Mikheev-Smirnov-Wolfenstein effect | Franz Vranitzky