Anyway after making sure it isn't a fluke and that it could really be a long term thing, I have been using headfit the past few days and have been tuning with more music instead of just the included test signals and I've come up with a .27ms setting instead of the .34 making the timbre more natural. It also so happens that the calculation for my head size should be around .284ms but the plugin skips from .27 to .29 and I find .27 less stressed sounding. I also compensated the ILD HF to make it so that the pink noise bleeding to the opposite side doesn't sound unnatural. I got a -28db value there.

to tighten up the center image, I go to .34ms though but maintain the same settings on the other parameters

I still find your crossfeed to be too congested for my taste. I am not saying that it fails to simulate speakers so far as imaging is concerned more than other crossfeed plugins. All I am saying is that what I want is something more like the meier natural crossfeed over headrooms crossfeed implementation for example. I just want a deaf ear effect elimination with minimal coloring of the sound. I don't want to feel like my headphones are being toed in or put farther away from me. All I want is the same headphone sound with less fatigue. The thing about trying to simulate speakers with a crossfeed circuit is that the usual monitor type headphone for example has a frequency response targeted to sound natural without any plugins. So when a crossfeed plugin is introduced, the sound usually becomes either like a bloated speaker with decent imaging or a thin sounding speaker with decent imaging all depending on whether the crossfeed plugin adds or subtracts bass.

I was just listening to the files at the meier site here by the way http://www.meier-audio.homepage.t-online.de/crossfeed.htm and tried out the samples of crossfeed vs no crossfeed and tweaked headfit to get a naturalness similar to the medium setting channel test. It turns out my setting on headfit was a little too high(thin) and a setting of .34, .20, and a change to around 1400hz in the f-central value made the signal sound almost identical to meiers medium setting. I am really liking the effect.

Meier's advanced crossfeed is exactly what I do. Specifically, the improved 3-channel variant of his, as described on this page:
http://www.meier-audio.homepage.t-online.de/crossfeed.htm - near the keyword CORDA.
The only thing I did is decouple ITD from ILD tunables.
If you set the high feed to -<lots> dB and ITD to be ~300 us (slightly higher than the default - set head wider by 1.5 cm) you will get almost exactly the same behavior. (Different slope of the lowpass is possible - tweak the steepness as you wish.)
Head-fit can only approximate that as it feeds mono signal.

I think you haven't tried to tune the tunables to taste - the defaults are pretty hefty.
The only difference is that the ITD is constant and not smoothly decreasing with frequency like in passive crossfeeds.
I can add an IIR allpass filter to do exactly this. (in addition to the much better Lagrange-interpolated fractional delay.)
That will by design introduce phase distortion - but will allow changing ITD and better emulation of passive crossfeed circuits.

BS2b is doing the original "natural" crossfeed. Mine will do too if you set ITD to 0 -> spread angle 0 with ITD tune 0.
The knee might be a bit different than the 1st order Bessel filter - and has tunable steepness.