Inverted density in recons

I am trying to run Frealign with CTF correction. I start with uncorrected (black) cryoEM particles and an initial model with positive density. My WGH value is set to 0.07, so I would expect to get CTF correction.

The resulting model has negative density and is of poor quality. My question is:

* Should the models have negative density?
* Are there any glowing errors in my parameter file/frealign parameters?

Thanks,
Neil Voss

Parameter file:
Iteration 0:
C PSI THETA PHI SHX SHY MAG FILM DF1 DF2 ANGAST PRESA
1 36.29 121.37 76.23 -0.25 -0.31 10000. 1 21272.5 19350.6 -12.55 0.00 0.00
2 308.27 60.01 255.49 -1.21 -1.17 10000. 1 21272.5 19350.6 -12.55 0.00 0.00
3 84.71 132.35 83.18 0.79 -0.93 10000. 1 21272.5 19350.6 -12.55 0.00 0.00
Iteration 1:
1 36.29 120.11 76.23 0.16 -0.64 10000. 1 21272.5 19350.6 -12.55 72.63 72.63
2 308.27 58.88 256.69 -2.29 -2.56 10000. 1 21272.5 19350.6 -12.55 73.43 73.43
3 84.15 133.22 81.68 -0.48 -0.52 10000. 1 21272.5 19350.6 -12.55 73.46 73.46


Job file:
frealign.exe << EOF
I,0,F,F,F,F,0,T,F,F,0,T,1
197,0,2.640,0.07,0.00,100,70,5,10,0
1 1 1 1 1
1, 500
I
1.00,2.64,15.00,85.00,2.00,120.00,0.00,0.00
5.30,50.00,10.60,100.00,0.00
stack
match
params.iter001.par
outparams.par
shift.par
0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
volume
weights
odd
even
phasediffs
pointspread
EOF