AI-Assisted Post-Production: Enhancing Workflows with AI Interpolation and Object Detection

AI-assisted methods are redefining workflows in post-production, enabling faster turnaround and more precise automation without sacrificing creative control. Recent advances in deep learning have produced practical solutions for long-standing technical challenges such as frame rate conversion, slate detection for dailies, and restoration of missing or damaged frames. This paper presents technical details of three such applications: (1) the use of RIFE interpolation augmented by NVIDIA TensorRT for accurate AI-based frame rate conversion, (2) object detection using Ultralytics YOLOv11 for slate identification in dailies workflows, and (3) generative AI-based frame inference for restoration of archival film. Case studies-including the recent 4K restoration of Sunset Boulevard at USC-illustrate how targeted models trained on relevant datasets can accelerate creative processes in production and restoration while keeping human operators firmly in control.

Published

2025-10-13

Content type

Original Research

ISBN

978-1-61482-966-9