Re: Lots of prior art, surely.
Windows and its "metafiles" should be a good match
Having just skimmed '633, I agree that it's basically a glorified graphics metafile. Metafiles are considerably older than the WMF format, though; they go back at least as far as the ANSI/ISO Computer Graphics Metafile standard (1986), and probably farther.
Some people are claiming that '633 specifically requires executable content in the "external shape" (the file), but here's a key quote from the patent:
Capabilities are action methods, symbol methods, or any other functions that allow the generation of information required to produce a graphical image.
What the patent calls "action methods" are actions: render, handle mouse input, etc. The "symbol methods" are properties: bounding box and that sort of thing. But "any other functions that allow the generation of information" is much too broad - it clearly could describe any graphics metafile format.
More generally, '633 just says "hey, we're going to load some code, and it's going to expose some interfaces that lets our app render some graphics". As others have pointed out, this is a well-established OS feature added to a commonplace OO-programming example. The rationale given in the patent is unconvincing. This patent is bogus and I hope the court throws it out.