create_eog_epochs does not detect the maximum signal during blink events

[stfnrpplngr]

Hi,
using the mne.preprocessing.create_eog_epochs() function does result in weird solutions. The epochs center is not the center of the peak. I would guess it has something to do with the filtering during blink detection. What is your opinion?

Average EOG plotting channels Fp1 and Fp1

EOG epochs plotting channels Fp1 and Fp2

[agramfort]

yes it's not detected properly but it should not be a pb to estimate an SSP