[xcrysden] Determine the angstrom to pixel ratio for printed images

[? ??]

Dear all, 

I would like to print some .xsf structure files using XCrysDen script. My .xsf files are exactly of the same structure (cubic diamond), but have different orientations relative to the cartesian directions x, y, z. The problem is the printed images have different sizes of atoms. 


To summarize, my question is, 
Can I fix the angstrom/pixel value before printing? Alternatively, can I know the angstrom/pixel value after the printing (I do not want to do the calibration myself since there are many structure files)? 

Another solution to this problem is to understand how XCrysDen determines the angstrom/pixel at the beginning of opening a structure. Unfortunately, the src code is too complex for me to find out the answer. 

Actually, it is enough to determine the relative values of angstrom/pixel for different structure files. 

Libin