Why re-extracted particles have the same resolution as bin2 particles

I used bin2 particles to run NU refinement, which got 3.54A map, after I re-extracted particles, it’s still 3.54A, what’s wrong with that? i used Titan Krios G2 without energy filter.
I tried several times but sometimes even worse. Does anyone know the possible reasons?
Thanks a lot!

What was the pixel size after binning? If alignment wasn’t limited inherently by sampling, but rather particle heterogeneity, then increasing the sampling rate won’t make a difference to the outcome. Instead, you may have to consider addressing compositional/conformational heterogeneity.