Beware of Imshift!!

I just had a crazy experience.

I worked very hard to properly stack my images. I had a stacked H alpha and a stacked continuum. I needed to shift one so I could subtract them, so I used imshift with these parameters:

                                                           I R A F  
                                            Image Reduction and Analysis Facility
PACKAGE = immatch
   TASK = imshift

input   =            stack_w16  Input images to be fit
output  =               goaty2  Output images
xshift  =                 -0.8  Fractional pixel shift in x
yshift  =                  1.1  Fractional pixel shift in y
(shifts_=                     ) Text file containing shifts for each image
(interp_=                 sinc) Interpolant (nearest,linear,poly3,poly5,spline3,sinc,drizzle)
(boundar=              nearest) Boundary (constant,nearest,reflect,wrap)
(constan=                   0.) Constant for boundary extension
(mode   =                   ql)

And my noise (standard deviation of sky) QUINTUPLED!!! (Went from 3ish to 15ish). I tried used interp=linear instead and everything was fine. My noise stayed the same as it was originally. So do NOT use interp=sinc!!!

(this is a bit confusing because when you are running mscimage, i believe that the interpolation of choice is sinc. hmmmm. maybe i should go back and check if that is affecting noise.)

A bit later...
Ok, just checked the results of mscimage and they seem ok. Even though I was using the sinc interpolant, this procedure does not seem to increase the noise in the images.whew.