Digital Twins (DTs) have gained significant attention for integrating real-time sensor data with digital models, yet their application in structural dynamics is often limited to calibrating idealized design models using modal data. In this study, we explore whether the incorporation of as-built geometry impacts the dynamic model calibration process for a two-story steel-timber hybrid building on the Virginia Tech campus. A 3D point cloud (3DPC) of the building was captured using LiDAR scans, and the geometric dimensions of the structure were refined by cross-referencing the LiDAR data with design drawings, ensuring an accurate representation of the as-built structure while identifying geometric inconsistencies between design and construction. Challenges such as noisy and occluded point clusters were addressed using statistical outlier removal and filtering methods to enhance the representation of structural components. An as-built finite element model was developed based on this processed data, while an as-designed model was constructed from the original plans. Simultaneously, operational vibration data were collected using accelerometers mounted on each steel beam-column connection, enabling modal shapes and frequencies to be identified through operational modal analysis (OMA). Model calibration was then performed by incorporating the as-built geometry obtained from LiDAR data, updating the model to reflect actual geometric characteristics such as member centerline alignments. After these refinements, the calibrated model closely matched the OMA results, demonstrating the impact of as-built geometry on structural behavior. By integrating LiDAR-based geometric data and vibration-derived modal parameters, we bridge the gap between as-built geometry and dynamic model calibration, contributing to advancements in reality capture, model updating, and decision support for structural dynamics.