A dynamic image virtualization system and method configured to utilize an AI model in order to conduct a reduced latency real-time prediction process upon at least one input image, wherein said prediction process is designated to create free-viewpoint 3D extrapolated output dynamic images tailored in advance to the preferences or needs of a user and comprising more visual data than the at least one input image.