Raster and Vector File Formats
The advent of Information and Communication Technology (ICT) has had a profound influence on art and art forms. The computer pervades all spheres of work and leisure and promises to become the most ubiquitous of all things. In the field of graphics too, art, paintings, sketches and photographs are being not only digitized, manipulated and enhanced, but also created, produced and disseminated through a wide range of hardware and software applications. Computer graphics both generated and digitized from conventional media are now posted on websites and reproduced in print and the electronic media in different forms, sizes and shapes. A photograph taken by a digital camera may, for example, be blown up to hundred times its original size and put up on a hoarding, a drawing generated by a software application may be replicated and manipulated and collated again to produce animation. In logo creation a computer graphic may be used in a wide variety of publicity and promotional material both electronic and hard print. In all these processes of conversion, modification and reconversion, the prime emphasis is however on quality, or to be more specific the resolution of the graphics.
Different types of media require different quality of graphics. While a low resolution photograph may be suitable for a website, a big banner advertisement would call for a much better quality graphics. All graphic designers would therefore have to have a very sound knowledge of the factors that affect the quality of computer graphics, they have to know which formats could provide them the quality standards that they are looking for within the constraints of a particular media. To be able to do so, the first and foremost requirement is to understand the two basic formats of computer graphics – vector and raster. To put it very simply, vector images are composed of connected lines expressed in mathematical terms; whereas, raster images are a collection of connected dots. Scanning of original or source artwork such as the hard print of a painting with some photo editing, print or scanner software such as Corel PhotoPAINT or Adobe Photoshop produces a digitized raster graphic. Vector graphics, on the other hand, are created when a vector illustration program such as CorelDRAW or Adobe Illustrator are used to draw or paint. In other words, raster graphics are usually the digital reproduction of an original source art, whereas, vector graphics are computer-created graphics. Raster graphics can also be converted to vector formats by using conversion programs such as CorelTRACE or EuroVECTOR.
Each of these formats has its own distinct advantages and disadvantages in relation to each type of media. The graphics belonging to each format have their distinct properties in the way that they are generated and in the ways that they can be modified or manipulated. “These two formats are quite different from one another, yet they contrast and complement one another when used appropriately for the desired final output method.” (Busselle, 2004)
Vector graphics utilize geometric modeling of images. Geometrical standards such as lines, curves and polygons are used to represent images in mathematical terms. The word vector is a synonym for line, which essentially implies that the vector format works on the depiction of lines through mathematical interpretations. Every line, curve and node in a vector graphic is defined by the software by a mathematical description with help of fixed co-ordinates. “A vector graphics file may be a database of object definitions or a sequence of instructions that are ‘played’ by an interpreter. The latter are often referred to as metafiles and are generally designed to be interchange formats. An example will serve to make things clearer.” (Westcott, 2005)
Adobe Illustrator uses a mathematical system called Bezier math, named after French automotive engineer Pierre Bezier, to define and locate the lines and shapes in a vector drawing. The system uses line elements that pass through defined anchor points to create line illustrations. The software plots the basic shapes that are drawn by listing them as numerical co-ordinates in pairs of x and y. After the co-ordinates are defined, the program draws the line elements between the points plotted. The software stores instructions to stroke the outlines of the shapes, to fill them or to do both. An element called the control handle is used when the shapes being drawn become more complex. This element stores information on how a curve passes through a particular anchor point and to which co-ordinates it would lead to. This enables Illustrator to have mixed sections of curves and lines. “The pairs of x,y coordinates, combined with the mathematical controller for curves and such are together called vectors. These vectors are solid mathematical values that even an accountant would appreciate. They don’t change (but they can be modified easily).”(Lawler, 2002) Vector graphics are object-oriented. Vector fonts are created by connecting the nodes, lines and curves of the text objects. Each letter of the font is considered as a vector object. Thus in vector graphics a wireframe is create under the colors of the object. The skeleton that makes up the object is clothed by the colors.
Vector images comes in a wide variety of extensions including EPS or Encapsulated PostScript, WMF or Windows MetaFile, AI or Adobe Illustrator, CDR or CorelDraw, DXF from AutoCAD, SVG or Scalable Vector Graphics and PLT or Language Plot File from Hewlett Packard. All these extensions are based on the software application packages that generate the graphics, and each come with its own advantage and unique feature.
Vector images can be scaled up or down without any loss in resolution because they are defined in mathematical terms, and increasing or decreasing the size of a graphics translates into something as simple as multiplying the mathematical description with a suitable scaling factor. Since scaling can be done in vector graphics without any loss in quality, these graphics can be printed at any resolution that the printing device is capable of. This makes vector graphics as preferred for clip art as for very large bill board images.
Unlike raster graphics which are dependent on individual pixels or units of colors, vector images are made up of objects. Changing the color of these objects could be as simple as clicking inside the object and defining its color.
The file size in vector images is comparatively very small because vector files comprise long mathematical descriptions detailing each aspect of the image or object. A change in size would imply only a change in a particular numerical value. Even a change in shape would add only to the mathematical description which by itself would not take up much of space. Vector image file sizes are therefore inherently small.
Raster images are made up of a collection of pixels. The pixel is the smallest unit of the raster image. The two words ‘picture’ and ‘element’ come together to form the term ‘pixel’. A pixel in a computer image actually contains information on the color and tonality of the corresponding spot in the source image. Monochrome information, duotone information and color information in RGB, CMYK and other color palettes can be represented by each pixel. The information stored against each pixel displays what the computer interprets as the actual color of the corresponding tiny part of the source image. Raster images are also called bitmapped images because each image is composed of rows of bits or pixels mapped or arranged into a pattern to finally render the source image in digital form.
Raster in German means ‘to screen’, which is appropriate for raster images as they are screened as a grid of pixels. In computer graphics, raster images are used for analog photographic material or continuous-tone images that have been digitized by scanning. Raster images come with different extensions depending on the software application and compression method that have been used. These include, BMP or Windows Bitmap, PCX for Paintbrush, TIFF or Tag Interleave Format, JPEG or Joint Photographics Expert Group, GIF or Graphic Interface Format, PNG or Portable Network Graphics, PSD for Adobe Photoshop, and CPT for Corel PhotoPAINT.
Raster images are not scalable. A raster image can be enlarged only to a certain extent, beyond which the image will show jagged edges. This happens because the pixels constituting the raster images get blown up as the image size is increased, and the edges of these individual pixels produce the jagged outline. The resolution of raster images is measured in dots per inch or dpi, and the actual output from the image is dependant on the resolution of the image irrespective of the printing device. For example if the resolution of a raster image is 300 dpi and it is printed in a printing device of 2000 dpi, it will actually print at 300 dpi and not 2000 dpi.
A large number of colors are required to render an accurate reproduction of the original source image. The human eye is usually not able to differentiate between an original and scanned image or of an art source if the scanning is done at 24-bit color depth or at 16 million colors. However, if the scanning is using a palette of 256 colors it would not be possible to reproduce the original accurately as a smaller range of colors is available to choose from. To overcome this hurdle a process called dithering or approximating colors that are not available is used by scanners. Though dithering approximates the original color in the image there is a visible deterioration in quality with the appearance of a distinct dotted pattern. The complexity of the process increases manifold when one attempts to change colors in a raster image. In such a case, the software application must not only be able to isolate the color or the range of colors that is sought to me modified, but should also be able to substitute the original colors with the desired colors. The biggest disadvantage of raster images is therefore in the inherent difficulties in editing or manipulating them.
Compared to vector images, the file sizes of raster images are comparatively higher because in raster image the software application has to store information on the exact location and color of each and every pixel in the image. The higher the resolution and the greater the color depth of the raster image, the bigger is the file size. File size becomes a handicap in the case of raster images as big files are especially not suitable for websites, and are difficult to transfer over the Internet. Even today in the days of high-speed broadband connections, it is difficult to transfer files over 1 megabyte over the Net.
Going by the inherent characteristics of the two computer graphics file formats, it is advisable to use the vector format, whenever possible, in all type, line art and illustrations and use bitmap or raster graphics only for photos or images with complex or non-uniform shading.
The modern day computer artist however has to be able to work both with vector and raster images. The ideal page layout program would be basically a vector-based application which would be able to at the same time create, import display and print raster images side by side with the native vector images. Examples of dual usage can be found in application software such as Adobe Photoshop – an image editor – which incorporates vector based paths which can be exported as native vector files. On the other hand, major illustration programs such as Deneba Canvas, Adobe Illustrator, CorelDRAW and Macromedia Freehand allow bitmaps to be embedded in the vector files they create.
Even in very diverse applications such as in the case of mappings of an earth observatory, images would be a combination of vector and raster with “annotations and continental outlines in a vector format (EPS) and data in a raster format (TIFF). Each covering the same area and with identical aspect ratios.” (Simmon & Sutton, 1998)
A very interesting point to note is that almost all software applications dealing with graphics nowadays, including animation software and CADD applications, use vector format. Printers such as PostScript printers use vector graphics and PostScript fonts which are actually vector fonts. Since vector images are more scalable and malleable, converting rater to vector images lends then accessible to a greater degree of manipulation and wider range of application.
However, dot-matrix printers, laser printers and monitors are raster devices. This implies that all vector images have to be converted to raster before they can be printed out or displayed by the raster devices. Therefore, conversion from vector to raster and also from raster to vector is required under different circumstances and situations. Conversion from vector to raster is more easily accomplished than conversion from raster to vector.
From Vector to Raster
To ‘rasterize’ means to convert a vector-based image to a bitmapped image. During this process, the mathematical formulas and geometric objects which make up a vector image are converted into groups of dots, using a specific resolution. Rasterization algorithms differ according to the shape of the geometrical elements of the image that is being converted to vector graphics. A triangle, a line, a polygon and a circle will all have their own vectorization algorithms. Once a vector file has been rasterized as a bitmap image, the resolution can not be changed without affecting the quality of the image.
From Raster to Vector
Raster images are transformed to vector drawings through a process termed as Vectorization. As raster images are inherently imprecise, the final vector output could vary widely in defining the exact number of vectors and the precise position of the co-ordinates from one conversion effort to another. It is therefore essential to employ well-established mathematical methods to take the variations into consideration and achieve the most precise conversion possible. Different conversion algorithms achieve these essentially through three basic approaches:
1. Using automatic conversion software which traces the dots in the raster image and translates them to mathematical terms of a vector image. A large number of software is available for automating the process of conversion of raster images to vector formats. Adobe Photoshop and Illustrator both come with their plugins for vectorization.
2. Manual underlay conversion specifically in CAD in which the raster image is inserted in a CAD application and the vectors are drawn over it in CAD.
3. Manual Digitizing in which the mouse or pen of a digitizing table is used to manually trace over each original line of the raster image.
The manual methods are still utilized even though automatic conversion software are available because the automatic conversion software are not able to do a clean job, especially when the raster image itself is of low quality, and manual reworking is inevitably required.
One way of achieving better vectorization is to convert all colors in the raster image to grey scale prior to the vectorization. This enables the vectorization method to differentiate between pixels that belong to the main image and pixels that belong to the background.
Vectorization also resorts to a process called negation. In negation, all the colors in the image are first converted to their corresponding photo-negative color. This is particularly useful for the vectorization of a blueprint or other negative-based originals where the lines are lighter in color than the background. In the simple edge detection method of vectorization, a negative image can be more effectively used in the case of a black and white raster image where the outer edges instead of the inner edges of the objects can be obtained. The brightness of objects is also adjusted in raster to vector conversions because such adjustment changes some pixels to background intensity and vice versa.
Despeckling is a process is a process which produces an image which does not have clusters of pixels that distorts the image and results in an inaccurate vector. Despeckling cleans up dirty scanned images with dark speckles which are not a part of the image.
Vectorization finds a wide variety of applications. In paper-to-CAD conversion or drawing conversion, Computer-Aided Design (CAD) drawings such as blueprints are first scanned as raster images and then vectorized and stored as CAD files which can processed and manipulated.
In Geographic Information Systems, maps are created by vectorizing satellite or aerial images. Vectorized photographs can be easily integrated into geometric designs such as logos.
Every person who works with computer graphics therefore has to master the art of working both in raster and vector formats. Options for conversion from raster to vector and vise versa are available if the situation so warrants depending on the type of media that the work is to be displayed in.
Busselle, J., 2004, Raster Images versus Vector Images, [Online] Available. http://www.signindustry.com/computers/articles/2004-11-30-DASvector_v_raster.php3 [June 04, 2009]
Lawler, B., P., 2002, Vector that, Charlie. A quick study in file types, history, metaphors and apple pie, San Luis Obispo, California
Simmon, R., Sutton, M., 1998, Image Formats: Vector and Raster, NASA.
Westcott, K., 2005, Preservation Handbook, Vector Graphics, AHDS Preservation Handbook, Arts and Humanities Data Service.