The claimed Voynich page

[oshfdk]

Depending on my exact lines, I got ~9.5% height difference, and ~7.5% width difference. The exact numbers depended on the exact rotation and choice of where to draw the boundaries. A stray word, longer space or including downwards tails/flourishes somewhere skews the numbers slightly. Using the top loops of the p and the top crossbar of the p makes a difference etc etc 

But the copy is always clearly smaller than the VMS.

I was measuring the width difference of the text, ignoring the picture. I think if we assume that one document as a whole (ignoring the extra leaf and the roots) is a projection of the other, it should be easy to find the best transformation by marking a few points on both and solving some linear equations, but I'm not very experienced in this.