Sampling theory, pt 1: WPFE in pictures

I got a few questions and comments about my comment that WPFE "botched" the sampling rules for bitmaps/images. I tried to explain in the comments, but its clear I wasn't clear.

So, I'll post a clearer text/mathematical explanation shortly but here's the issue, by visual example.

First a simple "linework" image:


Next, the same image if I apply a 400% scale to the canvas, as rendered by WPFE (I cropped the result to the upper left corner):


And now, the same image, with a 400% scale as produced by my (new) rasterizer - this is also the same image that Flash 8+ and OpenGL will produce:


See the problem?

I'll explain why this isn't a black and white issue, but why I think MS landed on the wrong side of it shortly, in a follow-up. The problem is covered well in the seminal computer graphics memo from Alvy Ray Smith, "A Pixel is NOT a little square..."

Alvy Ray, of course (being Alvy Ray), is not wrong in his final conclusion, but that doesn't make his point right for all applications either, and therein lies the rub... most UI designers, I'll posit, think of images as exactly little bags of rectangles...