0xsline/StoryGen-Atelier

AI-assisted storyboard and video generation tool. Uses Gemini for generating storyboard text and frames, Vertex AI Veo for generating transition clips, and ffmpeg for stitching the final video. Built-in logs and gallery management.

What it solves

StoryGen Atelier automates the complex process of turning a story idea into a full video. It eliminates the manual effort of writing scripts, generating consistent storyboard frames, and creating smooth transitions between those frames to produce a coherent cinematic sequence.

How it works

The tool uses a three-phase automated pipeline:

  1. Transition Analysis: It uses Gemini to analyze pairs of storyboard images (Shot A and Shot B) and generate a specific "Transition Prompt" and duration to describe how to move from one to the other.
  2. Clip Generation: Using the prompts from the first step, it calls Vertex AI Veo (or MiniMax) to generate video clips that connect the start and end frames.
  3. Final Assembly: It uses FFmpeg to stitch these individual clips together into a single, lossless MP4 video file.

Who it’s for

Content creators, filmmakers, and storytellers who want to rapidly prototype videos or create AI-generated cinematic stories from text descriptions.

Highlights

  • End-to-End Workflow: Handles everything from script and frame generation to final video stitching.
  • Interpolation Chain Strategy: Uses a sliding window approach to ensure visual continuity between shots.
  • Multi-Model Support: Integrates Google Gemini and Vertex AI Veo, with optional support for MiniMax models.
  • Management Tools: Includes a built-in gallery for saving/loading projects and a logs dashboard for tracking generation history.

Related

  • Project
    thu-ml/Causal-Forcing

    A framework for high-quality real-time interactive video generation using Causal ODE or Causal Consistency Distillation to enable efficient few-step autoregressive diffusion.

  • Project
    Robbyant/lingbot-video

    An open-source MoE video generation model dedicated to embodied intelligence, designed to produce physically rational videos for robotics and physical world understanding.

  • Project
    NikoDemon80/ComfyUI-H3-Motion-Context

    A ComfyUI extension for MiniMax H3 that enables seamless chaining of video and audio clips by using latent-based continuity to prevent quality loss and motion and sound jumps.

  • Project
    huangserva/ComfyUI_MiniMaxH3_Director

    A collection of ComfyUI workflows for MiniMax H3 Director, enabling text-to-video, image-to-video, and advanced video-to-video editing with reference materials.