Dear Alan,
The 3D reconstruction is oversampled by 2x zero-padding. The original 2D 
experimental projection is not oversampled. The CTF (a continuous 
function itself) is calculated at the discrete, non-oversampled points 
of the 2D experimental projection.
HTH,
Sjors

On 08/09/2015 08:23 PM, Alan Pryor wrote:
> Hi Sjors (and fellow RELION enthusiasts),
>
> Based on your papers, RELION performs CTF correction as it is assembling the experimental projections onto an oversampled grid. Normally, one artifically oversamples projections by simple zero-padding. However, this introduces correlation between Fourier points, some of which have values that have been phase flipped by the CTF and others that have not. On the other hand, the calculated CTF is more accurate with larger arrays and suffers from fewer artifacts, so it would seem desireable to CTF correct with a large array size. In principle one could crop out a larger array from the raw micrographs, but then this image is not oversampled. How do you account for this?
>
> Thanks!
> -AJ

-- 
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres