The use of 3D imagery (In one form or another) can be found in almost every media-related industry in the world, from logo creation, to designing a billboard onto which even more 3D imagery could be placed. 3D modelling is perhaps most prominent in the entertainment industry, and in particular the gaming industry.
Recently, science has begun to use 3D models to better observe the natural world. Real-life objects that can be scanned and converted into a 3D model and allow greater observation of details. 3D modelling is also incredibly important in architecture, as it allows designers to view the interiors and exteriors of buildings within a 3D space, just as they would in reality. Film production, much like game production has shifted into the realm of 3D also, allowing more dynamic and interactive visual effects and gameplay elements, and raising the graphical quality immensely.
3D development software are all built differently with different purposes in mind. One program may be suited better for rendering, while the other animation, so on and vise versa. Choosing the correct program for you is quite important, depending on what path you aim to take.
Blender has the instant advantage of being freeware and available for anyone to use across any platform. However, being freeware it’s not going to be nearly as dedicated any certain aspect as much as 3Ds or Maya is, for example. I personally found that Blender’s’ UI isn’t as user friendly as Cinema 4D.
Cinema 4D (Among most other 3D development software) Isn’t free, but is commonly used with a student license, so it’s not completely off limits to everyone. Cinema 4D enables easy to use UI and rendering settings, along with various plug-ins to create a powerful modeling program. However, like Blender it’s not necessarily movie-grade and any professional might find themselves lacking various features.
Sculptris This is the first dedicated modeling program I used. It’s free, which means anyone dipping their toes into the world of 3D modeling can try it out. It’s a very dumbed-down version of Mudbox, and is only really applicable for modeling organic shapes, as any hard-surface modeling proves extremely difficult. It doesn’t come with any sort of renderer, but the models can be exported into Zbrush for further development.
This is a pre-render still from the opening cinematic in Halo 2: Anniversary. The model, texture, and rig is intended for a sequence with photo realistic graphical quality. This particular model exists in more than one format, the model has low-quality counterpart that can be rendered real-time and exists in the gameplay.
This is a screenshot from BeamNG, a driving simulator created in the Cryengine. These models have been produced to be rendered in real-time, but also to enable soft-body physics and model deformation. The model allows the engine to deform and reshape the polygons according to dynamic physical inputs, such as collision, inertia and momentum. Any model can have these properties applied to them, but these have been modeled specifically to create a realistic effect.
3D GEOMETRIC THEORY
Geometric theory explains the building blocks of any 3D model, how they interact with one another and how they come together to create a 3D model. The 3 most basic elements of a 3D model are the Edges, Vertices, and the Polygons. The vertex is a single point in the space. The addition of another vertex creates an edge, and the addition of one more vertex creates a simple triangular plane, or a single polygon. A polygon can only have a minimum of 3 vertexes and 3 edges, as a triangle does.
A model doesn’t necessarily have to consist of tris, quads (A polygon with 4 edges and Vertex) can be used, and is more often used to create shapes of large surface area or low geometrical complexity. Tris are more often used on character faces and organic geometry that requires more complexity, for the sake of detail or animation.
CONSTRAINTS OF 3D
As with most things, 3D isn’t perfect and a number of drawbacks and formalities have been recognised over the years, and it’s a 3D modellers’ job to overcome and work around these drawbacks…
Perhaps the most important aspects of 3D modeling is the time is takes to render your model. In any other case that isn’t gaming, you cannot afford to render your model in real-time simply due to the graphical limitations that enable real-time rendering. Before you even begin modeling you must make the decision; do I want to render in real-time, or not. This creates an interesting set of circumstances- for instance, low polygon models capable of being rendered in real-time have the advantage of rendering instantly but are always going to present a lower graphical quality to a model that has been rendering for 2 hours. It’s a fine balancing act.
Depending on your polygon-count, file sizes (particularly within animation) can be enormous and can take even longer to render. Finding available space to store the information can also prove to be difficult, and even more difficult to transport.
The same model, pre-rendered (Left) and rendered in real-time (Right). Notice the lack of reflective surface on the goggles, lower-quality shadow maps, and some objects not drawing shadows at all.
This is a render of an AT-ST Walker from Star Wars Battlefront III in the Half-Life 2 Source engine, in real time.
In this image, depth of field and a degree of anti aliasing have been used to give the illusion of a camera focus. The objects in this scene don’t appear to be drawing any shadows, either due to the general cloud cover or graphical limitations. An artificial sky light has been used to illuminate the environment, and extra lights positioned around the model to have it stand out against the dull, overcast lighting of the moor.
This is a generally simple pre-render of a non textured, low-polygon mammoth in a blank environment. The choice to use a low-polygon untextured model is undoubtedly an artistic choice, rather than a limitation.
Any sort of Hypernurb/Subdivision Surface is clearly void from this model to create the stylised appearance, as the individual polygons are clearly visible. It’s likely that a sphere primitive was manipulated by moving individual vertex and polygons using extrude and grab tools to sculpt the body, while Spline techniques were used to create such geometry as the tusks, trunk and tail. Additional primitives may have been used to create the legs and the tuft of hair on it’s head by creating them separately, and using a bridge polygon tool to merge the meshes with the main body. For where additional detail is required, the artist may have used a knife tool to create more polygons onto a surface and thus allow for more geometry.
An alternative technique used to create this model would include that of sculpting, of which the artist would mould, shape and contort a primitive shape into a desired model. This technique is often effective when modelling more organic shapes, such as that of animals or landscapes, or in this case a mammoth.
This is an in-game render of a Pelican dropship in Halo 3 on an engine built specifically for the game.
This top-down or birds-eye view of the model shows the model’s largest surface-area, and the majority of the texture details. Various shaders have been used to achieve the realistic look of the surfaces, mainly bump mapping and specular mapping. The combination of shaders layered on the textures give the look of extra geometry where there is infact none, this allows designers to ‘cheat’ more detail without adding extra polygons, and thus saving on the performance of the game. In rarer cases, developers may use a tessellation shader to achieve an even more realistic look.
Only one light source is presented here, and that is the sun. The textures and shaders on the model capture it perfectly.
This is a render of a planetary environment in Anteworld built on the Outerra engine.
Anteworld is still mostly an interactive tech-demo. It’s for rendering planet-sized environments, all the way down to a single blade of grass, in in real-time and without any loading screens. This scene quite clearly uses a form of global illumination, with atmospheric fog and mist. The games vehicles’ feature spot lights and point lighting when the environment darkens during it’s real-time day-night cycle. LODs play a gigantic part in this game, trees don’t need a draw distance since they are only flat textures, but small rock geometry and texture resolution fades and/or vanishes as long distances, to help the GPU and CPU cope.
3D rendering is the process using Open GL that involves converting a 3D mesh into a 2D image, often coupled with shaders to produce a desired style of image, for example; photorealism.
Direct X is the name for a group of APIs that are commonly used throughout game development, for platforms as early as the Sega Dreamcast. Direct X is a set of ‘instructions’ for the GPU. It is responsible for calculating the the hardware specifications of your platform, and setting up the rendering quality accordingly. Direct X 1,2,3 and so on are iterations and each one adds certain performance enhancements and flexibility.