i have the second two. i have a "vanilla" 1W Luxeon version (made by eGear) of the XPG K2 - so other than in brightness and narrowness of the hot spot (i'm guessing here as this is typically true of K2's), it, i.e. the eGear 1W Luxeon, is similar in operation from a UI and battery-power standpoint.
XPG K2
-------
i'm guessing based upon my experience with the somewhat similar eGear, that it is a decent headlamp. nothing spectacular, IMO. for the price of $30, i'd call it a good value. IMO, K2's are not the best choice for most headlamp applications due to the narrowness of the hotspot and beam. of course, optics could be used to deal with this, but it is a characteristic in most designs. i suspect that it is similar here in the XPG K2. the XPG K2 due to its greater brightness will have a shorter burntime than the eGear Luxeon version that i have.
Atom
-----
as T says, the Atom is the same as the Cyclops.
i own an Atom, owned a Cyclops (now a boy at Church owns it), and also own a small headlamp similar to both of these is one made by eGear. the Atom and the Cyclops are very easy to twist on and off by the outer bezel/lens unit (they both turn together). the eGear's main downside is that it is a little more difficult to turn b/c one needs to turn just the lens portion, and not the entire bezel. it's BIG upside is that instead of using 2xCR2016 cells, it uses 2xCR2032 cells!! so instead of ~90mAh of stored energy, one has ~225mAh of stored energy at the same voltage. very nice in terms of microlight burntime. the eGear is no brighter than the other two similar headlamps, so its burntime is longer.
now, i know what y'all are thinkin'. i tried it, but it just so happens that the battery receptacle in the Atom is *NOT* deep enough to accomodate two CR2032 cells and the threaded portion of the bezel/lens is not sufficient to allow it to "grab" the receptacle threading.
i also tried ONE CR2032 and the LED is MUCH dimmer. i also doubt that one will be able to make use of the greater stored energy of the single CR2032 due to the fact that the voltage is already too low to properly fwd. bias (Vf) the LED. so, as the voltage of the cell drops, this situation is only exacerbated. someone (TIN, IIRC) recently tried this with the Nichia DS equipped Photon Freedom Microlight and proved that for ~24h, IIRC, the 2xCR2016 powered test case was brighter and then approx. as bright as a 1xCR2032 powered test case.
in all three of these little headlamps, optics are used to focus the beam into a very distinct circular output pattern (similar to the Aurora Jupiter LED flashlights, if anyone has experience with these, but of course these three microlights are no where near as bright - an that's just it, they ARE just microlights - please, keep that fact in mind when deciding IF to purchase).
Fowlproof LED Headlamp
-----------------------
CONS: 1) head stick out ***WAY*** too far, 2) *NO* declination adjustment on the head - this is a MAJOR *CRAB*, IMO, 3) BRIGHT, but *NOT* as bright as one might think for a Mfr. claimed ~390lm of o.p. (output), 4) while switch is large enough to operate while wearing gloves, the feel on my particular unit leaves something to be desired
PROS: 1) detachable from headstrap for clipping to a stiff ballcap brim, 2) multi-level o.p. selection, 3) dual power source, viz. the included rechargeable Li-ion battery pack or CR123A cells, 4) both "home" and "auto" chargers - very nice, indeed, 5) Mfr. claimed battery life is decent given the o.p. that light can produce. i haven't tested burntime myself, but have the feeling that the Mfr. claim might be a tad optimistic, meaning that the claimed time is *NOT* down to 50% of initial brightness, but the more common (particularly among headlamps "specs") down to 1 or 2 lux (or even 0.25 lux).
bottom-line: i don't feel like i wasted my money, but i won't use this headlamp much. if i could pick out one "crab" and one "good-point" it would be: crab=lack-of-declination-adjustment, and good-point=brightness.
hope this info helps.
