FITS creating compressed image data

Question

I'm trying to write data in FITS format using compression. Here's what I tried:

#include <vector>
#include "fitsio.h"
#define DIM 100
int main(int argc, char *argv[]) {
    fitsfile *fptr;       /* pointer to the FITS file, defined in fitsio.h */
    std::vector<double> data(DIM*DIM);
    for (unsigned int i=0; i<data.size();i++) data[i]=i;
    int s=0, status;
    fits_create_file(&fptr, "!test3.fits", &s); status+=s;
    fits_set_compression_type(fptr, GZIP_2, &s); status+=s;
    std::vector<long> naxes(2); naxes[0]=DIM; naxes[1]=DIM;
    fits_create_img(fptr,  DOUBLE_IMG, naxes.size(), &naxes[0], &s); status+=s;
    fits_write_img(fptr, TDOUBLE, 1, data.size(), &data[0], &s); status+=s;
    fits_close_file(fptr, &status); status+=s;
    return status;
}

It works since return status is zero. But I can't open the file with anything. Moreover the file test3.fits returns

test3.fits: FITS image data, 16-bit, two's complement binary integer

While if I remove the line fits_set_compression_type ... I get

test3.fits: FITS image data, 64-bit, floating point, double precision

that is correct (and I can open the image)

What I do wrong here?

Answer 1

It looks like I create a valid fits file it's just imageJ not reading it properly. Downloaded ds9 at it works pretty well.

Answer 2

There are older FITS libraries that cannot deal with the FITS tile compression standard. Software based on recent cfitsio (and CCfits) and nom.tam.fits is generally compliant with that convention.