The CCD data frames given to CCDPACK can be of any non-complex HDS (SUN/92) numeric type (e.g. they could be of type _WORD or _UWORD - Fortran INTEGER*2, _REAL, _INTEGER or even _DOUBLE). CCDPACK usually processes the data using this type. On occasion, however, frames, such as the master flatfield, will not be returned in their original type. This is because normalising to a mean of one precludes data storage of a precision less than _REAL. However, the flatfield correction routine FLATCOR will return the data in your input type regardless of the flatfield type so types are preserved in the longer term.
If your input frames are of a mixed data type CCDPACK will preserve the data type of each individual frame. However, if you are combining mixed data types into a calibration master of some kind, CCDPACK will choose the least precise type which represents best all the input data types.
In the routines MAKEBIAS, MAKECAL and MAKEFLAT input images which have different physical sizes (because they have been previously sectioned, for some reason) will be padded to a common size before processing. This is so that no calibration data is lost.
The corrective routines (CALCOR, DEBIAS and FLATCOR) trim the data down to the size which contains the smallest dataset. The trimming processes occur separately for each input image. The most efficient method of processing is to keep the input data files of the same type and size, as this avoids costly trimming, padding, and mapping/unmapping of the data (CCDPACK always attempts to minimize the amount of re-mapping of calibration frames when processing lists of images).
The MAKEMOS application is specially designed to deal with datasets which may have very small regions in common and which produce large output mosaics.
CCDPACK