Benchmark: timing 10 iterations of convert, epeg, epeg_qx, gd, gtk2, imager, imlib2, magick, nconvert, tk...
   convert: 61 wallclock secs ( 0.00 usr  0.02 sys + 55.88 cusr  4.53 csys = 60.43 CPU) @ 500.00/s (n=10)
      epeg:  5 wallclock secs ( 5.18 usr +  0.12 sys =  5.30 CPU) @  1.89/s (n=10)
   epeg_qx:  5 wallclock secs ( 0.00 usr  0.01 sys +  4.33 cusr  0.15 csys =  4.49 CPU) @ 1000.00/s (n=10)
        gd: 48 wallclock secs (45.78 usr +  2.26 sys = 48.04 CPU) @  0.21/s (n=10)
      gtk2: 13 wallclock secs (12.53 usr +  0.85 sys = 13.38 CPU) @  0.75/s (n=10)
    imager: 21 wallclock secs (20.37 usr  0.98 sys +  0.01 cusr  0.10 csys = 21.46 CPU) @  0.47/s (n=10)
    imlib2: 11 wallclock secs ( 9.77 usr +  1.18 sys = 10.95 CPU) @  0.91/s (n=10)
    magick: 71 wallclock secs (66.87 usr +  4.29 sys = 71.16 CPU) @  0.14/s (n=10)
  nconvert: 47 wallclock secs ( 0.00 usr  0.01 sys + 45.54 cusr  1.17 csys = 46.72 CPU) @ 1000.00/s (n=10)
        tk: 16 wallclock secs (14.31 usr +  1.25 sys = 15.56 CPU) @  0.64/s (n=10)
         s/iter magick convert   gd nconvert imager   tk gtk2 imlib2 epeg epeg_qx
magick     7.12     --    -15% -32%     -34%   -70% -78% -81%   -85% -93%    -94%
convert    6.04    18%      -- -21%     -23%   -64% -74% -78%   -82% -91%    -93%
gd         4.80    48%     26%   --      -3%   -55% -68% -72%   -77% -89%    -91%
nconvert   4.67    52%     29%   3%       --   -54% -67% -71%   -77% -89%    -90%
imager     2.15   232%    182% 124%     118%     -- -27% -38%   -49% -75%    -79%
tk         1.56   357%    288% 209%     200%    38%   -- -14%   -30% -66%    -71%
gtk2       1.34   432%    352% 259%     249%    60%  16%   --   -18% -60%    -66%
imlib2     1.10   550%    452% 339%     327%    96%  42%  22%     -- -52%    -59%
epeg      0.530  1243%   1040% 806%     782%   305% 194% 152%   107%   --    -15%
epeg_qx   0.449  1485%   1246% 970%     941%   378% 247% 198%   144%  18%      --