So the answer to my needs is a replacement set of DT770's and a creative X3 instead of my G6?
All generic HRTFs using some average head shape model are more or less compromises for most people.
So customizing algorithm based on listener's head shape likely gives little or significantly improved positioning for most people.
(in case of Super X-Fi it needs photo of your face and left/right ear to tune HRTF)
Though of course for someone else resulting binaural simulation could well sound even worser than generic HRTF...
If game sounds had been developed even at 10% of the speed of graphics development, we would have had such customized HRTF working directly from true 3D audio data of the games for years.
That's actually what would have been possible with DirectSound before Microsoft killed it in Vista:
In DirectSound games sound cards were able to get sound source 3D locations around player character, sound samples and even rough acoustic model of environment and then used that to render chosen output channel configuration.
For comparison no matter how much they're hyped, 5.1/7.1 "surround" are very lossy and for start lack vertical data.
PC/games would need Dolby Atmos like sound format supporting vertical channels and also "sound objects" with meta data defining sound's location around listener to overcome that.