Video Regeneration Using Image Diffusion Model

In this paper, we propose a novel approach for video editing using a combination of image diffusion models and frame interpolation. Our method utilizes the custom image diffusion models to enhance or edit the video frames and generate high-quality outputs. We also incorporate frame interpolation techniques to improve the temporal resolution of the edited videos. Our approach involves applying the diffusion models to each video frame separately after sorting according to quality and speed we need and then using interpolation to create a smooth transition between the edited frames. Our experiments demonstrate that our method achieves faster with relatively less loss of quality results compared to existing video editing methods. We also evaluate the impact of different image diffusion models on the quality of the edited videos. Our work provides a new direction for video editing using image diffusion models and frame interpolation and can be applied to various video editing applications.