I think the Black Magic process (some sort of combination of laserdiscs release blending and/or film jitter extrapolation) extracts more information than a single frame.
But just to prove my point (this is just simple Lanczos resize, you could do better with deblocking filters and such).

Image 1
The screenshot (from above) from the X0 project

http://img168.imageshack.us/img168/3231/x0gn6.jpg

Image 2
The same screenshot enlarged to 133% (about the same enlargement you'd see going letterbox to anamorphic).

http://img168.imageshack.us/img168/8210/x0anamorphicvm4.jpg

Image 3
For the heck of it, (approx.) HD 720p version (with some edge enhancement.) (200%)

http://img234.imageshack.us/img234/5575/x0720pv2px6.jpg

Do you feel like the picture is terribly degraded? And this is WITHOUT access to their raw data (or specialized software) before they encoded, resized, jpeg'd or whatever-ed the image before posting it here.

Want more? Here's the full size HD version:

http://img217.imageshack.us/img217/5272/x0720pv2fullsh1.jpg

I'd watch that.

EDIT: True avisynth filters used for 720p sharpening snapshot... looks better too.