README.md

---
pretty_name: CineBoard3D++
license: cc-by-nc-4.0
language:
  - en
tags:
  - 3d
  - blender
  - storyboard
  - camera-planning
  - narrative-grounding
  - dynamic-scenes
  - look-before-move
  - arxiv:2606.26964
size_categories:
  - n<1K
configs:
  - config_name: story_index
    default: true
    data_files:
      - split: inventory
        path: metadata/stories.jsonl
---

# 🎬 CineBoard3D++: Dynamic 3D Story World Dataset

## 📊 Dataset Summary

CineBoard3D++ is a collection of editable, movie-inspired 3D story worlds built with StoryBlender for narrative-grounded camera planning and world visual attention. It brings together story scripts, animated characters, scene geometry, and shot-level configurations in native Blender projects.

The benchmark covers **50 stories, 457 scenes, 1,585 shots, and 3,197 3D assets** (836 plot-related and 2,361 supplementary assets). It supports evaluation of subject perception, intent consistency, and trajectory quality.

This dataset accompanies the paper **“Look-Before-Move: Narrative-Grounded World Visual Attention in Dynamic 3D Story Worlds.”**

## 📄 Paper

**Title:** Look-Before-Move: Narrative-Grounded World Visual Attention in Dynamic 3D Story Worlds

**Authors:** Jiaming Bian, Bingliang Li, Yuehao Wu, Pichao Wang, Zhi Wang, Hailan Ma, Huadong Mo†, Zhenhong Sun†

(†: Co-corresponding author)

**Overview:** Look-Before-Move treats the camera as an active observer: it determines what the narrative requires it to see before planning movement. A Semantic Observation Contract specifies visual requirements, Monte Carlo Viewpoint Search selects feasible viewpoints, and Semantic Trajectory Grounding connects them into coherent, collision-aware camera motion.

## 🔗 Project Page

- 🌐 **Project:** [Look-Before-Move](https://engineeringai-lab.github.io/Look-Before-Move/)
- 📜 **arXiv:** [2606.26964](https://arxiv.org/abs/2606.26964)
- 📑 **PDF:** [Read the paper](https://arxiv.org/pdf/2606.26964)
- 💻 **Code:** [EngineeringAI-LAB/Look-Before-Move](https://github.com/EngineeringAI-LAB/Look-Before-Move)

## 📁 Dataset Structure

Each story is packaged as a single archive under `stories/`, such as `12 Years a Slave.tar`, `A Beautiful Mind.tar`, and `The Godfather.tar`. Story packages contain:

- 🎞️ **Blender projects** (`.blend`) with editable scenes and character animations.
- 📝 **Story scripts and JSON metadata** describing assets, layouts, scenes, and shots.
- 🧩 **Project resources**, including 3D models, textures, and animation assets.

The [story index](metadata/stories.jsonl) lists archive paths, project entry files, and checksums.

## 🚀 Usage

1. ⬇️ Download a story archive from [`stories/`](https://huggingface.co/datasets/EngineeringAI-LAB/CineBoard3D-plus/tree/main/stories).
2. 📦 Extract the archive, keeping its project folders together.
3. 🖥️ Open the story’s `.blend` file in Blender (usually `step_9.blend`).
4. 🎨 Explore scenes and animations, edit the world, or use the [code](https://github.com/EngineeringAI-LAB/Look-Before-Move) for camera-planning experiments.

## ⚖️ License

CineBoard3D++ is released under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/). Third-party assets retain their applicable licenses and attribution requirements; see [LICENSE.md](LICENSE.md).

## 📚 Citation

If you use CineBoard3D++ in your research, please cite:

```bibtex
@misc{bian2026lookbeforemovenarrativegroundedworldvisual,
  title={Look-Before-Move: Narrative-Grounded World Visual Attention in Dynamic 3D Story Worlds},
  author={Jiaming Bian and Bingliang Li and Yuehao Wu and Pichao Wang and Zhi Wang and Hailan Ma and Huadong Mo and Zhenhong Sun},
  year={2026},
  eprint={2606.26964},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url={https://arxiv.org/abs/2606.26964}
}
```