High performance images is a complex subject partly because of the apparent fragmentation in the browser ecosystem. It is useful to examine the history of images on the Web and specifically the history of the HTML
Images, the way we know them on the Web, almost didn’t happen. The
<img> tag is nearly ubiquitous in web development, so it is no surprise that nearly all modern application platforms and documents have followed in its path—supporting the same formats and styles. It wasn’t that images weren’t conceived of when Sir Tim Berners-Lee invented the Web; in fact, he used images and diagrams as part of his memo, proposing hypertext and the 18 elements. Inline images, however, were not one of the elements or use cases he described. Instead, the purpose of hypertext was to provide text and then link to documents. Those documents could be other hypertext files or binary files such as PostScript. It was assumed that the user would navigate to an image and not embed it into the text.
At the time, CERN was using NeXTSTEP, so it is not too surprising that PostScript and Encapsulated PostScript (EPS) were the primary image formats that the first-generation browsers could render natively in the browser application. This was mostly because NeXTSTEP provided handy APIs to render this content. Still, the expectation was that the user would select a hyperlink and the image would load in a separate ...