# IsoStage 2D

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Noureddine Noureddine](https://cerebralvalley.ai/u/DevFireTv)
- **GitHub:** https://github.com/DivProGamez/IsoStage-2D
- **Demo video:** https://youtu.be/yPpZKhNr9Ls
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4821

**IsoStage 2D — One Image. A World You Can Walk Through.**

IsoStage 2D is an AI-assisted tool that transforms a single illustrated room into an editable, playable 2D game scene.

Game developers and digital artists often have beautiful environment artwork but must manually define walkable areas, furniture collisions, player spawn points, and engine-ready scene geometry. This process is time-consuming, especially with isometric illustrations where parts of the floor are hidden behind objects.

IsoStage combines Meta Muse Spark 1.3 for semantic scene understanding and ground-contact estimation with Meta SAM 3.1 for object segmentation. It converts their outputs into proposed walkable floor regions and furniture collision footprints that developers can inspect, correct, and save.

Users can then upload their own character sprite, explore the original artwork using local collision physics, and export the resulting scene to a Godot 4 project.

Our working Magic Room demonstration includes genuine recorded Meta inference, human-reviewed geometry, playable character movement, collision testing, scene persistence, and a Godot export tested in Godot 4.7.2.

IsoStage does not promise perfect automatic geometry. Human review is an intentional part of the workflow, allowing developers to correct uncertainty while retaining control over their game worlds.

## More from Meta Global AI Developer Hackathon

- [To The Moon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4818)
- [Yopido Adaptive AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4819)
- [Invarith](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4820)
- [Muse Team](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4822)
- [DepaPranata](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4823)
- [opc](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4824)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4821. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt