Working with resampled data

Hi,

I downsampled my particle stack by a factor of 2, using the resample.exe program, and refined my map up to the point where the FSC is high up to the Nyquist frequency of this downsampled dataset. My refinement resolution limit is at a much lower resolution to avoid overfitting.

Then, I performed reconstruction using the original stack, and because the resample.exe program performs cropping in Fourier space, I expected my FSC to be good up to at least the highest resolution shell of the downsampled dataset, but it's actually worse than that. How can that be explained?

Thanks for helping!