AI Turns Sitcom Nostalgia into Interactive 3D Models
What if you could reimagine your favorite TV set as a fully navigable 3D world with just a photo and a prompt? That’s exactly what happened this week when a Twitter user threw down a challenge to two AI models: create Monica’s apartment from Friends, starting with nothing more than a set photo.
From Set Photo to Floor Plan in Seconds
The first prompt went to Banana Pro, an AI model designed for image understanding and spatial reasoning. The request: "Give me the floor plan of Monica's apartment from TV show Friends based on this set photo." Banana Pro quickly whipped up a full apartment layout, identifying room boundaries, doors, and even the iconic living room window. No need for hand-drawing or architectural know-how. The model infers spatial relationships and outputs a blueprint-style plan.
Jumping to 3D with Multimodal Models
But the real flex came next. The user passed Banana Pro’s floor plan to Opus 4.5, instructing it to create a 3D model of the apartment using Three.js, a popular JavaScript 3D rendering library. Opus 4.5 took the plan and generated a full 3D environment, complete with walls, doorways, and recognizable furniture placements. The result: a digital space straight out of Central Perk, ready for exploration or remixing.
Why This Is a Big Deal
This workflow showcases the growing power of multimodal AI. Models can now move seamlessly from image to structured floor plan to interactive 3D scene, all by chaining together specialized systems with simple prompts. What used to take hours of manual modeling or scene recreation is now possible in minutes, no Blender skills required.
It’s not just about sitcoms. Architects, game designers, and real estate pros could soon use similar AI workflows to turn quick sketches or reference photos into editable 3D assets, shrinking project timelines and lowering creative barriers.
Takeaways for Builders
- Chaining image-to-plan and plan-to-3D models unlocks rapid prototyping for digital environments.
- If you’re building creative workflows, consider how multiple models can hand off data for richer outputs.
- The gap between media nostalgia and interactive experiences is closing fast, powered by generative AI.
- Want to try it? Play with Banana Pro, Three.js, or explore the latest multimodal LLMs.
With AI bridging the gap from photo to floor plan to 3D world, your favorite TV settings are just a prompt away from becoming your next creative playground.
Vision dreamt up by @scottstts
