This research objective focuses on improving the fidelity, realism, and applicability of interactive traffic simulations through parallel testing, where identical scenarios are examined both in real‑world conditions and in photorealistic virtual environments. The overarching goal is to identify and quantify the key factors that influence the credibility of virtual traffic simulations, thereby reducing the dependence on costly and time‑consuming on‑road testing.The dataset integrates multiple data sources originating from the U SMART ZONE test polygon in Ústí nad Labem and supplementary urban locations, including measurements of vehicle speeds, traffic flow composition, origin–destination patterns, and behavioural observations of drivers in complex traffic situations. These real‑world data inputs are combined with detailed 3D models, high‑resolution photogrammetric reconstructions, and dynamically parameterized simulation scenarios used in the virtual environment.Methodologically, the data support the development and iterative refinement of simulation scenes, the validation of virtual models against real‑world behavioural outcomes, and the calibration of dynamic actors (vehicles, pedestrians, cyclists) for use in training and validation of ADAS/AD systems. The dataset also underpins the creation of a VR‑based educational tool designed to simulate pedestrian‑vehicle interactions for training and public‑awareness purposes.The resulting multi‑layer dataset enables traffic engineers, researchers, and ADAS/AD developers to conduct sophisticated analyses of driver behaviour, scenario criticality, environment perception, and algorithmic robustness. It is intended to support research reproducibility and foster broader use of virtual testing in transport engineering, safety research, and simulation‑based education.