Computational heart modeling is a promising approach for improving the prognosis of patients born with congenital heart defects. To create accurate physics-based digital cardiac twins of this population, it is crucial to accurately represent the highly diverse and unique subject-specific heart geometry. In young pediatric patients, this is a challenging endeavor given the lack of high-spatial-resolution imaging data and the risk of slice misalignment. In this study, we set up a multistep shape morphing and slice correction approach to accommodate these challenges and establish a population of biventricular heart models for a variety of healthy, Fallot, and Fontan pediatric patients.