Authors: Wavhal Vedant Hemant, Prof. S. S. Bhor, Prof. M. C. Kolse, Prof. S. K. Said
Abstract: Artificial intelligence and machine learning are reshaping visual effects pipelines in modern cinema. Traditional VFX requires extensive manual labour for repetitive tasks such as rotoscoping, chroma keying, rigging and matchmoving. This paper reviews AI-driven techniques that automate these processes. It examines video object segmentation and high-resolution video matting that generate mattes without a green screen; structure-from-motion and visual SLAM that recover camera paths for matchmoving; markerless motion capture based on parametric body models and large motion-capture archives; neural rendering with neural radiance fields and real-time Gaussian splatting for novel-view synthesis and environment reconstruction; learned fluid simulation that approximates expensive physical solvers; and generative video synthesis and facial reenactment used for digital de aging and performance transfer. The paper discusses how these methods improve efficiency and reduce post-production time and cost, and then examines the accompanying concerns of dataset copyright, performers' likeness and consent, data privacy and workforce displacement, together with content provenance standards as part of an ethical framework. A five-stage AI-assisted VFX pipeline covering acquisition, automated preparation, asset creation, rendering and review is proposed, with research gaps, objectives, a comparison of tasks and challenges. The review concludes that AI should function as a collaborative assistant that extends rather than replaces human creativity.
