Gait analysis is crucial in Parkinson's disease (PD) diagnosis to clinically observe and quantify abnormal motor patterns. A primary gait biomarker is the center of mass trajectory (CoM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> ) that express the global coordination of forces, neuromotor commands and musculoskeletical poses. Also, from such trajectories is possible to analyze main locomotion moments (LM), such as forefoot, midfoot and heel strike, commonly altered in PD. However, the CoM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> requires additional devices, e.g. force platforms, limited to only a few steps or markers associated in video analysis but altering the natural motion gesture and losing description of LM. This work introduces a markerless approach to compute the CoM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> followed by a Lagrangian global magnification. Additionally, a magnified video reconstruction allows to better observe gait patterns, useful for medical observation analysis. The Evaluation was performed on a control (7 patients) and PD (7 patients) video set, achieving a proper LM description w.r.t raw CoM <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> captured in videos.