HaineiFRDM: Structure-Preserving Diffusion for Film Restoration under Fast Motion and Diverse Defects

The HaineiFRDM model has been introduced for film restoration, addressing challenges in fast motion and diverse defects. It employs a patch-wise strategy with position-aware global fusion modules to preserve scene structure and enhance texture consistency through a frequency-based module, achieving high-resolution restoration on a single 24GB-VRAM GPU. This model is significant for practitioners as it improves restoration quality and reduces memory requirements, facilitating better handling of film defects in dynamic scenes.