Researchers from Nanyang Technological University Revolutionize Diffusion-based Video Generation with FreeInit: A Novel AI Approach to Overcome Temporal Inconsistencies in Diffusion Models