Cryo-electron microscopy (cryo-EM) single-particle analysis (SPA) has revolutionized structural biology, enabling the efficient determination of structures at atomic or near-atomic resolutions. However, a common challenge arises from the severe imbalance among various conformations of the vitrified particles, leading to low-resolution reconstructions in rare conformations due to a lack of particle images in these quasi-stable states. To address this limitation, we introduce CryoTRANS, a method that enhances the resolution of rare conformations. CryoTRANS constructs a quality-preserved trajectory between density maps of varying resolutions, utilizing an ordinary differential equation parameterized by a deep neural network. This ensures the retention of detailed structures from high-resolution density maps. Leveraging one high-resolution density map, CryoTRANS significantly improves the reconstruction of rare conformations, demonstrating its potential through real-world cryo-EM cases and providing valuable insights into dynamic biological processes.