End-to-end high-fidelity avatar reconstruction using 3D Gaussian splatting with monocular video as the sole input | Synapse