Texturing Essay

“To make a texture believable, you have to be able to convey to viewers exactly what the surface would feel like if they were to reach out and touch it.”

-          Leigh Van Der Byl, 2004, pg.3

In order to bring a 3D model to life as it were, it needs to have a surface and texture which would make it believable as a real world model. 3D programmes have created a range of dynamic capabilities when it comes to texturing surfaces of 3D objects. In this essay, I will explore 2D and 3D texture mapping techniques in terms of how they are each utilised in 3D software procedures and their comparative differences. I will also highlight the advantages and disadvantages of each of these methods.

I will firstly define and explore 2D texture mapping techniques. Hong Zhang defines 2D texturing as the mapping of a “2D image to the surface of a 3D object.” (2006, pg. 409). This effectively allows for photographic images taken from real world spaces to be projected on any object within a 3D space. The precise mapping of these 2D images onto 3D objects involves a technique called UV mapping. This involves the process of ‘pinning’ points of a 2D image to specified points on the 3D object, depending on how the user wishes the image to be projected. Peter Ratner notes the advantages of UV mapping, saying it “works well for irregular shapes” and “is used for precise placement” (2003, pg. 215).

3D software further provides the user with different methods (presets of sorts) of mapping these image projections depending on the shape of the 3D object to be textured. These are now highlighted and explained:

1.       Planar Projections – These are used on flat surfaces (such as walls and floors) and can be compared to “projecting a slide through a slide projector.” (Ratner, P. 2003, pg216).

2.       Cylindrical Projections – Such projections are used to wrap around a texture, “similar to a bark around a tree trunk” (Ratner, P. 2003, pg216). These projections are useful for objects like poles, sticks etc.

3.       Spherical Projections – These projections ‘roll’ around the object “as if you were wrapping skin around a ball” (Ratner, P. 2003, pg216). Rounded objects like balls, planets and light bulbs use this method.

4.       Cubic Projections – Cubic projections wrap around 6 sided objects. These objects can be anything from a fridge to a CD case.

In addition to these methods, users define upon which axis/axes the projection/s will be implemented. In a 3D space these are the x, y and z axes. The combination of the different projection methods coupled with the ability to use them in any variation of the axes makes for very specific and accurate mapping capabilities.

By contrast, 3D texture mapping does not work on 2D based images. Instead, as Isaac Victor Kerlow defines, 3D texture maps are “solid textures that exist on the surface of an object as well as inside the object” (2000, pg.255). These textures are generated by algorithms performed by the computer and are referred to as “procedural texture maps” (2000, pg.255). Procedural as a term comes from computer science jargon which serves to “distinguish entities that are described by program code rather than by data structures” (Ebert, David S. 2003, pg12).  These mathematical functions performed by the computer render abstract and seemingly random images which then occur throughout 3D object to which they are applied.

 Such a technique is useful for 3D objects which have a recurring, yet random, pattern which pervades the entire object, for example, marble or wood. They also become useful in environment simulation and have indeed been used since the earliest days of 3D. Ebert notes that early 3D practitioners like Schacter and Ahuja used a 3D texture generator known as Fourier synthesis to “generate texture imagery for flight simulators” (2003, pg.11).

Comparatively, each of these techniques (2D texturing versus 3D texturing) has both its advantages and disadvantages.

Let us explore some of the main advantages of 2D texturing. Firstly, 2D textures can be practically anything which is a photograph/painting/drawing/bitmap image and this allows for a very realistic appearance on the 3D object (if mapped effectively). 2D texturing also provides a quicker rendering of the image (even though the image file would generally be larger than a 3D textured one). Also, due to the very specific nature of UV mapping, the user is able to define the exact parts of an object he/she wishes the image to be projected upon. Ratner also notes that 2D texturing is most common in the industry. (2003, pg.215).

On the other hand, there are also notable disadvantages. First and foremost, 2D texturing does not occupy a 3D space but rather requires “specific mapping coordinates in order to be rendered correctly” (Autodesk, 2006, pg.444). This means that if the user were to make a cross-section of an object to which they had applied a 2D image, only the surface would have the texture. Furthermore, 2D images applied to 3D models tend to look much flatter than a 3D generated texture. There is, of course the option of bump mapping which “simulates the appearance of a rough surface”, but even this technique can only provide a simulation of depth (Ratner, P. 2003, pg.221). 2D images also have the problem of becoming more and more pixellated the closer one zooms into them; this is due to the fact that they are fixed images of a fixed resolution. In addition, as Brian Ross notes, there can be problems of stretch marks, seams and tiling (if a small image is applied repeatedly over a large area.) (1998, pg.1).

Turning to 3D texturing, there are also notable advantages and disadvantages.

The first advantage of 3D texturing is that “the procedural representation is extremely compact” (Ebert, David S. 2003, pg.14). While it may take more time for the computer to render out the 3D texture, it is in fact a very small file (usually kilobytes) in comparison with a 2D image (usually in megabytes) (Ebert, David S. 2003, pg.14). Secondly, due to it being based in mathematical formulae, it has no fixed resolution and it will therefore remain fully detailed regardless of how close one zooms in on it. (Ebert, David S. 2003, pg.14). The mathematical formulae provide 3D texturing with the additional advantage of having infinite variations and therefore avoiding any seams or ‘tiling’ effects (Ross, B. 1998, pg.1). Ross further notes that such 3D generated textures have the additional option of being animatable.

In terms of disadvantages, it is noted that to properly generate a 3D texture can be difficult and often requires complex programming. (Ebert, David S. 2003, pg14). In conjunction with this, it is often easier, and more accurate, to use a found image which accurately represents the texture instead of trying to generate it. Furthermore, although 3D texture files are smaller than 2D image files, they often take more time for the computer to evaluate and render out than a 2D image would. Lastly, aliasing and anti-aliasing can be tricky and “is less likely to be taken care of automatically than it is in image-based texturing” (Ebert, David S. 2003, pg.15).

In conclusion, this essay has comparatively defined and analysed the techniques of both 2D and 3D texture mapping. It has looked at how both provide unique tools and approaches to creating believable textured objects, how each technique suits specific texture mapping purposes and finally it has made a comparative list of each method’s advantages and disadvantages.

(1150 words)

Works Cited:

1.      Autodesk. 3DS Max 9 Essentials. Autodesk, Canada. 2006

2.      Ebert, David S. Texturing and Modeling: A Procedural Approach. 3^rd Ed. Morgan Kaufmann Publishers. San Francisco, CA. 2003.

3.      Kerlow, Isaac Victor. The Art of 3D Computer Animation and Imaging. 2^nd Ed. John Wiley & Sons. New York, NY. 2000.

4.      Ratner, Peter. 3D Human Modeling and Animation. 2^nd Ed. John Wiley & Sons. Hoboken, New Jersey. 2003.

5.      Ross, Brian. Texture Mapping. http://www.cosc.brocku.ca/Offerings/3P98/course/lectures/texture/. 1998. Web.

6.      Van Der Byl, Leigh. Lightwave 3D 8 Texturing. Wordware Publishing. Plano, Texas. 2004.

7.      Zhang, Hong. Computer Graphics Using Java 2D and 3D. Pearson Education. New Jersey. 2006.

Introduction to 3D Software - Research Essay_01

3D Animation – Research Essay

Dean Barrett

Due Date: 7 March 2011

Topic: Discuss the different kinds of modelling techniques used in a 3D programme (Polygons, NURBS and Subdivisions) as well as the strengths and weaknesses of these methods.

“Three-dimensional modelling can be compared to sculpting”

-          Peter Ratner, 2004, pg 4.

Modelling in a 3D program presents the user with many different techniques and approaches to creating a model. This essay shall explore the properties and applications of three of these techniques or approaches: using and manipulating polygons, NURBS, and subdivision surfaces. It shall further discuss both the advantages and disadvantages that each of these techniques presents the user.

Let us firstly define and discuss the polygon. As Ratner defines it, “a polygon is a portion of a plane bounded by three or more lines or segments” (Ratner, P. 2004, pg 4). A polygon therefore contains a collection of points to which lines (or edges) connect  in order to form faces (or sides). Faces and points on polygons can then be manipulated in a number of ways to form the shape of what it is the user wishes to create. For example, the face of a polygon can be extruded (or even pulled back) to increase the length of the polygon or to add a certain detail. Points on the polygon edges can be moved, welded together and lengthened amongst many other possibilities. Due to their straight edges, polygons are useful in creating geometric shapes/figures which incorporate angular parts e.g. a table.

The advantages of using polygons are that they are generally very flexible and easy to manipulate. One can rotate them, join/split them, intersect them amongst many other possibilities which allows the user to create complex meshes from single primitives. The main disadvantage of polygons is that, due to their straight and hard edges, they have a very segmented appearance which George Avgerakis describes as looking “boxy and artificial” (Avgerakis,G. 2004, pg233). To overcome this, one can further subdivide these polygons “until the edges [sic] smooth out and disappear” (Avgerakis,G. 2004, pg233). This, however, presents another disadvantage associated with polygons and that is that in creating a more complex mesh to smoothen out one’s polygon shape, the polygon becomes a much more render intensive object which will inevitably take a substantially longer amount of time to fully render.

In order to overcome excessive polygon meshes, one can also make use of splines – and more specifically for this essay – NURBS. A spline, firstly, is defined as a “flexible line segment defined by edit points called vertices” (Ratner, P. 2004, pg 4). NURBS, an abbreviation for Non-uniform rational B-spline, are a type of spline which is “characterised by a set of control vertices (CV’s)” and which makes the creation of smooth surfaces and curves possible (Ratner, P. 2004, pg 4). NURBS are therefore mostly used in the modelling of organic objects (e.g. flowers).

The first advantage of using NURBS is that users can avoid the complex polygon arrangements used to achieve smooth edges and surfaces which allows for a smoother looking object with a lower render time. Tom Meade notes that NURBS use fewer control points which “translates into smaller file sizes” and, due to this, is a more efficient use of one’s computer’s RAM. (Meade, T. 2007, pg134). He also notes that it is easier to map textures onto these surfaces due to the “corresponding coordinate systems” between UV directions on a NURBS and XY coordinates on, say, a Photoshopped bitmap image. (Meade, T. 2007, pg135). Lastly, he notes that NURBS have a very accurate “conversion ability” from a NURBS surface to a polygon surface which also produces a very neat polygon geometry. (Meade, T. 2007, pg136).

The first disadvantage of using NURBS is the occurrence of seams. These occur, as Meade explains, due to the fact that NURBS surfaces are generally made from “multiple individual surfaces” (Meade, T. 2007, pg136). This results in one being able to see the spaces between different NURBS surfaces. These seams may be ‘patched’ together in a variety of ways but, once again, this increases the render time of the object. Another disadvantage is known as branching which refers to “parts of models that extend, or branch off, from the main surface or a group of surfaces” (Meade, T. 2007, pg137). In this scenario, “five-point junctions” become evident which then need to be modified because NURBS surfaces “can only contain a quad layout” (Meade, T. 2007, pg137). Lastly, while Meade also listed it as an advantage, he also notes that texture mapping can be a disadvantage of NURBS too, especially across multiple surfaces due to textures being applied to a variety of uneven surfaces, further complicated by the fact that UVs “cannot be edited separately from the mesh” (Meade, T. 2007, pg137).

Subdivision surface modelling was developed almost as a meeting point for polygons and NURBS. As Peter Ratner defines it, “this method uses a low polygon control mesh that applies a smoothing algorithm to bend the edges of polygons, giving them a curved appearance.” (Ratner, P. 2004, pg 5). Subdivision surfaces are useful for creating organic shapes from single surfaces and polygons.

The advantage, then, of a subdivision surface is that the user can achieve the smoothness of NURBS with the versatility of polygons and still avoid an excessive polygon count. The disadvantages of subdivision surfaces include, according to Dariush Derakhshani, requiring “even more computation than NURBS, and keeping models in subdivisions will cost you a lot of memory.” (Derakhshani, D. 2010, pg110). Due to them existing between polygon surfaces and NURBS surfaces, they are also difficult to apply UV texturing upon.

In conclusion, this essay has defined and discussed three of the modelling techniques dominantly used in 3D software today – polygons, NURBS and subdivision surfaces. It has explored the properties of each of these, their specific applications in modelling as well as looking at the advantages and disadvantages of all three. It has shown that, due to the wide variety of methods available to use when approaching modelling, it is useful to plan and map one’s ideas with regards to the implementation of these techniques in order to achieve the best result with the least amount of strain on the computer and, indeed, with the most efficient working strategy.

(1009 words)

References:

1.      Avgerakis, George. Digital Animation Bible: Creating Professional Animation with 3Ds Max, Lightwave, and Maya. 2004. The McGraw-Hill Companies, Inc. USA.

2.      Derakhshani, Dariush. Introducing Maya 2011. 2010. Wiley Publishing. Indianapolis, Indiana.

3.      Meade, Tom. Maya 8: The Complete Reference. 2007. The McGraw-Hill Companies, Inc. USA.

4.      Ratner, Peter. Mastering 3D Animation. 2004. 2^nd Ed. Allworth Press, New York.

Concrete Concerto - Instrumental I wrote last year

http://soundcloud.com/playfan/concrete-concerto

Howdy all,
I wrote and recorded this little piece on Cubase AI4 mid-last year for a film project called Time in Concrete. My Cubase has a limited library so not all instruments (especially guitars) are of the timbre I want them, but I think musically it worked to the film's brief - very atmospheric,rhythmic but still relatively simple and unobtrusive over the picture.
Anyways, let me know what ya think...