In a small experimental project called "Yonkers Rises," a user tested Claude Fable 5.2 Medium by asking the model to generate a 3D city animation. The user first admitted they did not know where Yonkers was, so they checked its location and shape on Google Maps. To help embed that geography in their mind and to exercise the model's 3D modeling and animation capacities, they issued the following prompt:
Make a three js animation where yonkers stands up as a big mech (cool stand up motion) and átomos around manhattan and queens and Brooklyn, use openstreetmap mapping data
The prompt combined two technical elements: three.js for browser-based 3D animation and OpenStreetMap as the open-source mapping data to render city outlines. It also included a mixed-language/typo element: the word "átomos" appears to be a Hungarian keyboard slip rather than an intended English word.
Claude Fable 5.2 Medium inferred the intended meaning correctly, recognizing that "átomos" was likely a typo for the English verb "stomps" — i.e., the user wanted Yonkers to "stomp" around Manhattan, Queens and Brooklyn in the animation.
Why this matters
- Technical summary: the request ties spatial data (OpenStreetMap) and a web 3D renderer (three.js) together, while relying on a large language model (Claude Fable 5.2) to translate a natural-language prompt into instructions suitable for 3D animation.
- Error correction in context: the model demonstrated contextual understanding by correcting a language/typing error, which is useful when prompts include mixed languages or rapid, imperfect input.
- Potential uses: workflows like this can support urban visualization, game prototyping, educational simulations, and creative experiments where real-world map data is expressed as dynamic, narrative visuals.
The example does not state whether a finished three.js animation was produced or how polished any resulting model was; the key takeaway is that contemporary language models can assist not only with text generation but also with interpreting spatial and 3D-oriented prompts and resolving user input errors.
Conclusion
"Yonkers Rises" illustrates how open mapping data, browser-based 3D rendering, and a modern language model can be combined through a single creative prompt. The fact that Claude Fable 5.2 Medium corrected the typo underscores that such models are increasingly capable of handling messy, real-world user input.
