Browsing by Author "Mohammad Abdul Rehman"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Artio (Stable Diffusion Based Video Generator
    (UMT.Lahore, 2024) Mohammad Abdul Rehman; Muhammad Ashbil Bin Shahid; Muhammad Salman
    This work proposes iterative refinement and checkpoint merging for stable video generation. In addition, different diffusion models are used to remove artifacts and enhance video resolution. Early video generation systems (pre-2010s) emphasized joint processing owing to the deficiency of the quality of videos for analysis. Stable diffusion came as a breakthrough for image generation. However, extending the approach to videos is not straightforward. Videos should not only be generated but also appear realistic and temporally coherent. That is, the generated frames should be perceptually consistent and should constitute a logical sequence. Furthermore, we’re using real-world images to show how effective our model could work on real-time images