These graphs are probably deceptive because of their normalization: the peak output of the light seems to be rescaled so that it ends up at ~4000 "intensity counts". If we consider that for a given power, an HID bulb will produce a whole lot more lumens, and in the specific situation that we're talking about, the claims are something like 2.5x the output of halogen bulbs using less power, the graph starts to smell funny.
Here's a quote from wikipedia that might shed some light (sorry) on the issue at hand:
"[...] luminous flux is adjusted to reflect the varying sensitivity of the human eye to different wavelengths of light."
and then
"The lumen can be thought of casually as a measure of the total "amount" of visible light in some defined beam or angle, or emitted from some source. The number of candelas or lumens from a source also depends on its spectrum, via the nominal response of the human eye as represented in the luminosity function."
both from
http://en.wikipedia.org/wiki/Lumen_(unit)
so if we made generated graphs of the different headlight options in question and did not normalize for output, they would look look very different.