Powder bed fusion using a laser beam (PBF-LB) [1] enables geometrical design freedom to build parts for optimized functionality. Furthermore, PBF-LB allows microstructural design freedom. By controlling the solidification behavior microstructural adaptions can be made to obtain the full potential of the material. As the solidification rates and the thermal gradient depend on the local part geometry, new data-driven approaches, e.g. machine learning (ML), seem to be suitable for local microstructural adaptions. In this work an evaluation concept to analyze the thermal melt pool characteristics based on a high-speed camera is developed. The thermal radiation intensity of the melt pool is used to derive the thermal gradient and combined with an image rate of 41,000 fps the solidification rate is derived. The developed approach provides local data of the solidification for ML-based process adaptions but also serves for part individual quality assurance tasks.