The means of providing and using high quality textual information on the Web has received a great deal of attention. Of course, there is still plenty of low-grade textual information out there. But the requirements are well understood and the infrastructure is in place for those who wish to generate or use quality text.
Much graphical material on the Web is also of poor quality, but in contrast the requirements for high quality graphics on the Web have received much less attention. Granted, many graphics are mere adornment, but some are not. Graphics can be valid information in their own right.
It would also be dangerous and limiting to assume that browsing constitutes the totality of present and future Web use. In some applications, the graphical component can be essential. Thus, it makes sense to evaluate the particular requirements for quality Web graphics.
A sample of existing and potential application areas dependent on high quality images:
This paper will survey some major areas affecting 2 dimensional, static, raster graphics quality. Vector graphics (such as CGM), 3 dimensional graphics (such as VRML), animations and movies are not considered.
It will focus on areas such as accurate colour and tonal rendition which may as yet be unfamiliar to developers of Web user agents. Topics which are important for Web graphics in general, but do not strictly relate to graphics quality - such as compression efficiency - are also outside the scope of this paper.
Computer Graphics can be seen as a somewhat specialised area - particularly where accurate rendition is required. Yet parts of it need to be understood to produce good graphics. Experience has shown that users can be harshly critical of browsers with poor image quality, even if these same browsers excel in other areas. The thrust of this paper is to pull together separate strands of computer graphics as they relate to Web user agents, to make it easier to provide the increasingly high levels of graphical quality that users are demanding.
A number of different quality requirements are discussed in this paper. Attempting to implement a single one of these in isolation can be difficult, as there are strong interactions between them. Partial implementation still requires an awareness of the other factors. This paper highlights known interdependencies to make the implementors job easier and prevent nasty surprises.