If you assume people are lazy, and that they reuse the same file names for lots of things, it makes it easy to spot the pattern in less then 15 minutes

There doesn't seem to be any truely downloadable formats of the videos, but here's a start at getting the URLs for the text containers that identify where the streams are for a WMV capable player...

#!/usr/bin/perl -l

use strict;
use warnings;
use LWP::Simple;

my $base = "http://sportsillustrated.cnn.com";
my $path = "swimsuit/2007/02/03";
my $vidbase = "http://wmscnn.stream.aol.com.edgestreams.net/cnnsi";

my $vid_index = get "$base/features/2007_swimsuit/video/";

while ($vid_index =~ m{(/features/2007_swimsuit/video/.*?.html)}g) {
    my $page = get "$base/$1";
    if ($page and $page =~ m{/video/$path/(.*?)/include\.js}) {
    my $code = $1;
    print "$vidbase/$path/$code/video.ws.wmv";
    }
}
[download]