Browsing by Author "Asad Abbas"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    DreamBoard AI
    (UMT.Lahore, 2025) Muhammad Awais; Asad Abbas; Muhammad Hamza
    DreamBoard AI is a web based tool that helps filmmakers, animators and other creatives turn written ideas into full storyboards in minutes instead of hours. Using large language model (LLM) “scene planners” and Flux based image engines, the system writes a structured plan for each scene and then generates four matching frames while keeping characters, lighting and style consistent. A Python “Consistency Engine” checks every new frame with YOLO v8, CLIP and DeepFace to be sure the main characters still look the same from shot to shot. Users can edit, in paint, regenerate single frames or export the final board as PDF, PNG or ZIP. Real time WebSocket updates, JWT secured accounts and a PostgreSQL + AWS S3 backend make the app fast, safe and scalable. Internal tests show the first image appears in under 25 seconds and overall character matching accuracy reaches about 89 percent. By automating repetitive drawing tasks, DreamBoard AI lowers pre production costs and lets storytellers focus on creativity.