I think the big problem is none of them have got the balance right between offering more capabilities that are actually useful vs. how intrusive they are.
Its a VT100 (+colors) compatible terminal emulator (pyte) hacked to render into an HTML document, displayed using Webkit (pywebkitgtk). Currently requires python2.7, gtk, webkitgtk (only tested on Ubuntu 11.04 so far). Html output is done by using escape sequences to mark beginning and end of html strings or base64 encoded resources (js, images ...). You can even use js frameworks (used jQuery + d3.js to generate the circular tree in the 2nd screenshot). Communication between the terminal emulator and the client can be achieved by using console.log messages or some kind of HTTP (over stdin/stdout). Would love to get some feedback.
When I read this, the things that came to mind were elinks, lynx, w3m and the browser that lives in emacs. But maybe that's not what you mean - could you distinguish what you're describing from these?
w3m can display images in a X terminal (rxvt, xterm). It renders html as terminal text too of course. I suppose you could imagine rendering the whole thing (including fonts and text and whatnot) in a PixBuf and then display that in the X term but I fail to see why you would do that instead of having it done directly on X.
