3D Report – Understand theory and applications of 3D

3D software that is available and how it is used within the industry.

Some of the software that is available within the Games industry is:

Maya

3DS MAX

Softimage

This software is used for polygon modelling, manipulating individual and small groups of polygons these are good for making a base models that are then added to in different programmes.

Mudbox

ZBrush

These are sculpting types of programmes, these use a huge amount of polygons compared to the ones above, these are usually used when you want there to be a lot of detail within a character or an object.

Blender

Sculptris (this uses voxels)

--

In practice the software would typically be used like this; after the designs have been done and dusted, they would start is the programme Maya – they use this software to build the basic model of whatever it is that they are creating, be it a character or an object like a car.

They would then import that model into the programme ZBrush –this is to get the higher dependability of detail into the character or object – this is by the use of an amazing amount of polygons that will be used due to wanting the higher quality within the project that is being worked on.

 They would then Bake this into a normal or bump map, these both serve the same purpose as to what it is that they do, and also into ambient occlusion maps, this is where the software tries to approximate the way that light would give off of an object in real life. These are both to help the model when it is formed to look more realistic within its environment.

The baked textures will then be applied back into the Maya software and then this will be applied to the model.

They do this because it uses far less polygons, so the rendering time will be faster. So they make a high quality model and they then basically make that 2D and wrap it around the less polygon model in Maya, so that it looks to have all the detail, but in actuality there are a small number of polygons that the baking allows them to do.

Geometric theory

A polygon is usually a triangle, this is because there are only three points and no matter what point you are at you can always touch the other points and it will look like a triangle if there are more points used, the points in space can become distorted and what might have started out as a square has been worked out to look like more of an hour glass shape. However it is hard to create objects with triangles, it is much easier to do so with squares. These are later then converted into triangles, and this is easily done when building with squares.

Triangles can be represented by three points in space along the x,y,z axis. 3D modelling is basically manipulating these points into what it is that you want to create. These points are called a vertex. A vertex contains information about the x,y,z coordinates of where it is within the 3D space. It contains the UV coordinates as well, this is when doing texture mapping and is basically the same point but in 2D space rather than 3D, this is because UV’s are a 2D representation of a 3D model. It will also hold the information of the normal.

 A normal is an object such as line or a vector – this is perpendicular to a given object. This can be looked into at 2D and 3D level – in the case of 3D, the surface normal, or what I have been calling it, simply, normal, to a surface at a point P is a vector that is perpendicular to the tangent plane to the surface at P. 

Mesh construction techniques

Box /Subdivision Modelling

Box modelling is a polygonal modelling technique in which the creator would start with a geometric primitive, with is a cube, sphere, cylinder, etc – they then model this shape, refining it until the final design is achieved.

People that box model usually work in stages, starting with the low resolution model. They then refine the shape and then subdividing the mesh to smooth out hard edges and to add in different bits of detail. This method is then repeated until the model contains enough detail that it conveys the original design.

This is one of the most common forms of polygon modelling and is often used in combination with edge modelling techniques.

Extrusion Modelling

This is a method where the user creates a 2D shape which traces the outline of an object, possibly from a drawing. The user would then create a second image of the subject but from a different angle and would extrude the 2D shape into 3D, following the shapes out line. This is usually common for creating faces and heads. When doing this, in general an artist will model half a head and then would mirror it in some way; this will ensure that the model is symmetrical.

Edge/Contour Modelling

This is also a polygon technique; however this is fundamentally different from its box modelling counter part. Edge modelling, instead of starting with a primitive shape (cube, sphere, etc-) and then refining. A model like this is essentially built piece by piece, and this is done by placing loops of polygon faces along prominent contours, and then filling any gaps between them.

This is a good way to create a face; this is because it is harder to create one through box modelling alone. The precision of contour modelling can be invaluable. Instead of trying to shape a part of the face with a cube, which can be confusing, it’s much easier to build an out line, like, of a mouth or eye and then model the rest from that.

Digital Sculpting

This is a way of modelling that helps free modellers from the painstaking constraints of topology and edge flow, this is a type of modelling that allows them to intuitively create 3D models from scratch, in a similar fashion to sculpting digital clay.

This way, meshes are created organically, using a graphics tablet to mould and shape, almost like a sculptor would use utensils on clay. This has taken creating creatures and characters to a whole new level. This makes the process faster, and also allows designers to work with high-resolution meshes that may contain millions of polygons. They produce a previously unthinkable amount of surface detail in the models that are created.

Procedural Modelling

This is modelling that is generated algorithmically, rather than being created by hand. This is scenes or objects that are created based on user definable rules or parameters.

This is plugging in information about a scene into a computer programme, so imagine that there was a large amount of trees that you would like to create. Then using this algorithm I would be able to create number of randomized trees based on a premise. This would make a forest or a field or something along those lines.   

Displaying 3D objects

Direct3d from direct made by Microsoft –

“The Direct3D API imposes some constraints on the processing model in order to achieve optimal rendering performance. Direct3D 11 introduces the Compute Shader as a way to access this computational capability without so many constraints. It opens the door to operations on more general data-structures than just arrays, and to new classes of algorithms as well. Key features include: communication of data between threads, and a rich set of primitives for random access and streaming I/O operations. These features enable faster and simpler implementations of techniques already in use, such as imaging and post-processing effects, and also open up new techniques that become feasible on Direct3D 11–class hardware.” http://www.microsoft.com/en-gb/download/details.aspx?id=23803

Direxct3D is used to render three dimensional graphics in applications where performance is important, an example of this would be games. Direct3d allows applications to run full screen instead of embedded in a window, though they could still run through a window of programmed for that feature. It uses hardware acceleration if it is available on the graphics card; this is to allow the hardware acceleration for the entire 3D rendering.

OpenGL –

OpenGL (open graphics library) is a cross language, multiplatform application programming interface (API) for rendering 2D and also 3D computer graphics, this is used to achieve hardware-accelerated rendering. OpenGL is also profited by a non-profit company.

There are also a number of other game engines that allow for the displaying of 3D objects.

3D Rad
Ardor3D
Axiom Engine
Cafu Engine
Crystal Space
Grit

There are also many more game engines that could be mentioned, but there are too many to go into detail about in this report.

Level of Detail (LoD)

This involves decreasing the complexity of a 3D model representation as it moves away from the viewer or according to the metrics like, an objects importance or position. LoD techniques increase the efficiency of rendering by decreasing the workload at other stages, like the vertex transformations. When the Level of Detail is decreased it usually will go unnoticed because of the object appearance when in the distance or if it is moving fast.

Recently, LoD techniques also included shader management to keep control of the pixel complexity.

This is an image of a Rabbit going down in the amount of polygons, loosing some of its detail.

This is an image showing the object in the distance and in the foreground, as you can see, you do not really need all the detail you have in the creature in the foreground and you do in the background.

Pre-Rendered vs. Real-time Rendering -

Pre-rendering is the process in which footage is not rendered in real-time. Instead, the video is a recording of footage that has previously rendered on a different programme. Pre-rendered material may also be outsourced by the developer to an outside production company.

Real-time rendering is one of the interactive areas of computer graphics; this means that mock images are created fast enough that a player or viewer can interact with the environment through a computer. Video games are the most common place that real-time rendering is found. The rate at which images are shown is measured in frames per second. The frame rate is the measurement of how quickly an imaging device produces unique constructive images.

An advantage of Pre to Real is the ability to use graphic models that are more complex and more intensive. This is due to the possibility of using multiple computers over an extended period of time to render the end result.

A disadvantage of Pre to Real, in the case of gaming, is the lower level of interactivity with the player. This means that while these Pre-rendered scenes are playing on a game, a player can usually just watch, or do the very minimal in any instant.

Another disadvantage for Pre-rendered assets is that changes cannot be made during gameplay. A game that has Pre-rendered backgrounds is forced to use the fixed camera angles, also a game with a scene like this, generally cannot reflect any changes the game characters might have undergone previously in the game. 

 

Lighting

Real-time rendering

To make a model have a more realistic appearance one or more light sources are usually used in the scene when manipulating the model. However, this stage cannot be reached without completing the 3D scene being transformed into the view space, the view space

Pre-rendered

This type of rendering in a game has a problem where lighting cannot be easily change the state of the lighting in a convincing manner.

Pre-rendered lighting is a technique that is losing popularity. Instead processor-intensive ray tracing algorithms can be used during a game’s production to generate light textures, which are then simply applied on top of the usual hand drawn textures.

 Shaders

Are a set of instructions that are programmed into a computer to generate how light would bounce off a surface.

Real-time rendering

This is going to put a lot more on the shader software – this is because there are many more surfaces and angle in which light will bounce off, mostly in games, this is because a user can look around most environments from nearly any position, and this will change the angles of where light will hit a surface.

Pre-rendered

Shaders are put under less stress when they are programmed into Pre-rendered scenes, this is because they are fixed camera angles, so there are only so many surfaces that light will bounce off.

Textures

Simply put – Pre-rendered textures are higher resolution and have a lot more detail in them, compared to that of real-time rendered textures. This is because real-time rendering does not have the memory or power to be running high resolution textured assets and environments per-frame. Pre-rendered scenes can allow for the size due to it being more of a video than actual game play, so they can put the detail and resolution to use and make the pre-rendered scenes better.

Polygon count

Real-time rendering

In real-time rendering the polygon count is low, as to having to be able to render out a scene pretty quickly, and the frame rate has to be quite quick so that the environment and the character move pretty seamlessly though a scene.

The use of low poly meshes are mostly confined to computer games and other software that an outside user must manipulate 3D objects in real time because processing power is limited on personal computers or games consoles and these have a need for higher frame rates.

Pre-rendered

This uses a massive amount of polygons compared to real-time rendering. This is because it will already be rendered previously, so when this is in a scene it will run more smoothly than the real-time rendering.

Films or still images have a higher polygon count because rendering does not need to be done in real time, which would require higher frame rates. This type of rendering is usually not constricted by computer possessing power because they are usually use a large network of computers, which is sometimes known as a rendering farm. These are more detailed and higher quality because of the amount of polygons used.

This is the end of my 3D Report.