Stacking process across NaNs
Regardless of whether you use a mute, stacking should exclude both NaNs and zeros.
As a general rule, NaNs are always excluded. Zeros are excluded if they are at the far offset edge of the data (i.e. there are no non-zero, non-NaN samples between this one and the furthest offset).
Note: The steps on how to perform the stacking process can be obtained from the Stacking section.