While local galaxies fall into two dominant populations (passive, red, pressure-supported spheroids, and blue, star-forming disks), these familiar Hubble-type classifications do not apply as readily to high-redshift galaxies, the most massive of which are compact and red. Logically, these high-z galaxies likely become the elliptical population at z=0, but they must grow by 3-5 times in size in the interim. It is proposed that the size evolution of galaxies either occurs by accretion of smaller galaxies, whereby the compact core remains, or by adiabatic expansion due to mass-loss winds, whereby the entire galaxy expands. The key to distinguishing between these two scenarios is the accurate measurement of the size-mass relation. This requires sufficient resolution to measure effective radii and Sersic indices of the most compact galaxies over a wide field of view, at rest-frame optical wavelengths to avoid bias due to small-scale localised star-for mation. In this talk I will describe our project using the Gemini South Adaptive Optics Imager, with its unique capability of diffraction-limited near-infrared wide-field imaging, to image the cores of galaxy clusters over the redshift interval 1 < z < 2. I will present our results from the first galaxy cluster, as well as our methodology for processing the complex data from this new instrument (including successful correction for the quasi-static off-axis distortion, varying PSF and image ghosting).