Question about sound signature

The exact sound signature is hard to pinpoint because there is more to it than the graphs such as how up-front or laid back a headphone is.  Some of that information can be extrapolated from the frequency response especially toward the treble, but it is going to nail it down exactly from the graphs.  Another issue is found in the impedance.  A lot of dynamic headphones have tricky impedance curves and power management demands.  They are not typically solely resistive loads and the ability to drive them well can be tricky, but the sensitivity can reveal a little about that as well.

Tyll can probably speak more about the graphs, but I will give it my best shot.

Frequency Response:

The UE600 certainly go deeper into the bass and have less roll off than the HF3.  This typically leads to the lower octaves being heard a little easier, but it will depend on your source material if you hear it or even if it is there.  The HF3 have a +4dB hump at the 2 kHz mark which will be audible.  This is heard on a piano as the B and C at the far right of the keyboard.  The UE600 have a peak around 10 kHz which is almost +6dB above the HF3.  The 10 kHz area is where my ear starts to fine peaks in the FR curve damaging if they go to 0dB or above, which the UE600 does not so you should be fine there.

500 Hz Square Wave

A perfect reproduction of the 500 Hz Square Wave would be a nice very flat square wave, but because most headphones typically limit one end of the FR curve the wave becomes an estimate.  When a square wave goes into a headphones it will test multiple frequencies of the square wave.  For the 500 Hz square wave this would be 500 Hz, 1000 Hz, 1500 Hz, etc.  These are very audible regions of the spectrum which is why the frequency was chosen.  See how the HF3 has a large spike near the left side of each square wave.  To me this means that is going to have a louder representation of the fundamental ( 500 Hz ) than the rest of the multiples.  The immediate dip shows that it may lose some frequency components passed the fundamental.

To me both look similar enough to be called analytical ( flat frequency response ), but perhaps because the HF3 dips down below 0dB from 90 Hz on back is why some list is as "Too Analytical".

Another thing to consider with IEM headphones is how they seal, what filters if any do they use, how large is the sound aperture, and how is the resonator built.  All of those affect the sound as well.

A frequency graph will give you some information on how something might sound, but it is by no means a 100% accurate test to show sound signature.  Sound signature will vary from person to person.  To add, your ears will be different than the ones on the dummy that was used to test those headphones (and get the FR graph).  Added even deeper is the fact that IEMs are extremely reliant on the seal created.  One step further, tip selection also makes a difference.  A different tip can actually create a much different sound in some IEMs.  That being said, take everything I'm about to say with a grain of salt:

• Lows: The lows on the 600s will have more of a sub-bass presence along with a much larger punch in general.  A larger (but not as large) body will also allow better bass presance. 
• Mids: The etymotics will have much better midrange presance in general.  The may be livelier, but that is hard to discern.  From the looks, the Etymotics look more energetic in the higher instrumentals.  This energy may or may not also apply to the vocals.
• Highs: Complete toss-up really.  From the looks of it, the 600s may sound more open and slightly larger (soundstage wise due to the V-shaped created with the really high highs and bass).  Slightly more sparkle than the Etymotic. 

That's really all I can read off of the graphs.  Again, nothing will ever beat getting the two together and testing for yourself.  None of my observations I've made can be claimed for 100% accuracy (I'll be lucky to even get 75-80 :p).  Even with my attempts in reading the graph, I am still unclear on a few areas.  As you can see, this really isn't an accurate method of getting an actual sound signature out of a pair of headphones.