In sheet metal forming, the interaction between virtual models and the real world remains challenging. Process simulations can exhibit significant errors, and reliable measurements are often scarce during early production stages. This study presents a hybrid twin framework that systematically unifies computer-aided design, simulation, and measurement data in an adaptive manner. Central to this framework is a reverse engineering algorithm that reconstructs and transforms the geometry of deep-drawn components from optical scan data into B-spline surfaces. The algorithm demonstrated high precision, indicating its suitability for process control and geometric analysis. The hybrid twin framework integrates virtual data from simulations and real-world data, as evidenced by a sensor concept for inline surface measurement. The framework ensures robust and redundant measurement concepts by estimating complete geometries from a few systematically preselected measuring points. This adaptive approach permits continuous updates and extensions to the database, accommodating both sparse inline signals and offline inspection data. This framework provides a conceptual model for integrating direct feedback interactions between virtual and physical environments, thereby enhancing the precision of analytical and predictive models in sheet metal forming processes.