Posts

Showing posts with the label simulation

#4660 Google Genie Sims for Lunar Rover Training

#4660 Google Genie Sims for Lunar Rover Training #4660  Provide 15 useful Google Genie prompts for lunar rover training on the moon Here are 15 structured Google Genie prompts designed to generate interactive 3D training environments for lunar rovers. To maximize Genie’s action-controllable world model, each prompt includes the Environment Setup (lighting, geology, hazards), the Agent / Vehicle (rover type and mechanics), and the Control / Perspective perspective. 1. Permanently Shadowed Region (PSR) Ice Sampling World Prompt: Ultra-realistic, pitch-black interior of Shackleton Crater at the Lunar South Pole. Sharp spotlight illumination mounted on the vehicle, casting long contrast shadows over fine, powdery regolith and glints of subterranean water ice. Character / Agent: A 4-wheeled solar-and-battery hybrid lunar rover with high-traction mesh wheels and a forward-mounted robotic sampling drill. Perspective & Controls: First-person driver/navigational camera feed with ...

#4243 Google Gemini Project Genie; +Lunar Surface Integrated Operations Sim; +Capture

Image
#4243 Google Gemini Project Genie; +Lunar Surface Integrated Operations Sim; +Capture #4243  Tell me about Google Project Genie Launched as an experimental research prototype in January 2026, Project Genie is a web-based "world model" from Google DeepMind. It allows you to create, explore, and play in interactive 3D virtual environments generated purely from text prompts or uploaded images. Unlike traditional video generators (like Veo or Sora) that produce a fixed clip, Project Genie generates a playable space where you control a character in real-time. How It Works The system is powered by Genie 3 , a foundation world model trained on tens of thousands of hours of gameplay videos. It works in tandem with other Google models: Nano Banana 2: Used for high-speed image generation and physics simulation. Gemini: Provides the multimodal reasoning to understand complex prompts. When you move or jump, the model doesn't follow pre-programmed code; it predicts what the nex...