Ben Scott <> writes:
>   Some of you may have seen xkcd #1110, "Click and Drag", from Wed 19
> Sep.  If not:
>   Some people have analyzed it.  If it was a conventional image, it
> would be 165888 x 79872 pixels (W x H).  As a 32-bit uncompressed
> bitmap, it would be almost 60 gigabytes.  Printed at 300 DPI, it would
> be 46 feet wide.  The xkcd page uses HTML "stitching" and a lot of
> blank fills to accomplish this in about 6 MB of images.

Yes. In addition to being more space-efficient, the particular technique
of sparse-tile-stitching in use on that comic also actually helps to
frustrate attempts at downloading the whole thing or otherwise
`seeing the big picture'.

I initially wrote a quick script to try brute-forcing the problem
on the computer+network rather than spending too many brain-cycles
reading the JavaScript code, and found it thwarted by the
unholy amount of hole-iness in the map: you can't just start
at the center, walk until you hit `the end' of the world,
and then jump to the next row or column and walk back toward
the opposite edge, because it's not a single landmass contiguous
in all directions.

So I gave the (slightly obfuscated) xkcd JS a more thorough reading,
revised my script, and downloaded all of the tiles using slightly-
less-brutish force. Haven't had time to stitch them all together
myself, yet.

On the whole, I find that comic to be an amazing adventure in
both moral amibuity and moral-ambiguity.

>   Someone created a zoomable version (like Google Maps):
>  Run the browser window full-screen,
> and it's impressive, despite being a black-and-white stick figure drawing.

Hm. If that site actually provides a distinct set of tiles per zoom-level,
that might make a neat easter-egg in FoxtrotGPS....

"Don't be afraid to ask (λf.((λx.xx) (λr.f(rr))))."

gnhlug-discuss mailing list

Reply via email to