Kepler and K2 data analysis reported in the literature is mostly based on aperture photometry .
Because of Kepler ’ s large , undersampled pixels and the presence of nearby sources , aperture photometry is not always the ideal way to obtain high-precision photometry and , because of this , the data set has not been fully exploited so far .
We present a new method that builds on our experience with undersampled HST images .
The method involves a point-spread function ( PSF ) neighbour-subtraction and was specifically developed to exploit the huge potential offered by the K2 “ super-stamps ” covering the core of dense star clusters .
Our test-bed targets were the NGC 2158 and M 35 regions observed during the K2 Campaign 0 .
We present our PSF modeling and demonstrate that , by using a high-angular-resolution input star list from the Asiago Schmidt telescope as the basis for PSF neighbour subtraction , we are able to reach magnitudes as faint as K _ { P } \simeq 24 with a photometric precision of 10 % over 6.5 hours , even in the densest regions .
At the bright end , our photometric precision reaches \sim 30 parts-per-million .
Our method leads to a considerable level of improvement at the faint magnitudes ( K _ { P } \gtrsim 15.5 ) with respect to the classical aperture photometry .
This improvement is more significant in crowded regions .
We also extracted raw light curves of \sim 60 000 stars and detrended them for systematic effects induced by spacecraft motion and other artifacts that harms K2 photometric precision .
We present a list of 2133 variables .