HomeLaunchesLucid
23

Lucid: Generative Simulations powered by Fast World Models

Unbounded video games and fully differentiable reinforcement learning gyms

Hey YC! We’re Alberto and Rami, the founders of Lucid.

https://youtu.be/fnoyvrGOwIA

We’re building generative simulations powered by fast world models. Instead of using traditional game engines with hard-coded physics, our models learn to simulate reality from pixels, enabling real-time interactive environments. With it we will train robots in their own imaginations and make unbounded gaming experiences. We trained the fastest world model ever seen to simulate minecraft end-to-end (20+fps on a gaming GPU).

The Problem: Game Worlds Are Static & Expensive to Build

Modern game development is slow, expensive, and constrained:

  • GTA V took 3 years, 1,000 employees, and $100M+ to build—AAA game budgets are skyrocketing and they’re not getting any better.
  • Despite the price tag, these games are inherently static, with predefined environments, objects, and interactions.
  • Players can’t truly shape the world—every door, street, and event is pre-scripted.

Meanwhile, robotics faces its own bottleneck—AI models trained in simulators (MuJoCo, Isaac Sim, Gazebo) fail to generalize to the real world (Sim2Real gap) because today’s simulations are hand-coded approximations of physics rather than learned from real-world data.

Our Solution: Generative World Models

Lucid replaces traditional game engines with a generative simulation engine that learns from data rather than being manually programmed.

  • Every frame is generated in real-time, conditioned on player actions.
  • Trained on video, not game scripts—our models learn the rules of physics directly from pixels rather than hardcoded logic.
  • Infinite, dynamic game worlds—players can generate and explore entirely new environments just from a text prompt or sample concept art.

A Neural Minecraft Simulator

We trained a neural network to simulate Minecraft end-to-end—every pixel is generated in real-time, learned from 200 hours of gameplay.

  • Runs at 20+ FPS on an NVIDIA 4090—5× faster than existing world models (Decart’s Oasis <4 FPS).
  • Aggressive latent compression—we utilize a VAE with 128x spatial compression allowing us to vastly reduce the amount of tokens needed to represent a single frame

What’s Next? Training on the Real World

We’re now training our models on real-world video data to build a general-purpose universe simulator for:

  • Gaming: The last game engine humanity ever needs—generating unique environments dynamically from simple text or multimodal prompts.
  • Robotics: Simulations that actually match reality—training embodied AI models in diverse, realistic environments. A fully differentiable, learned simulation framework for reinforcement learning.

Want to learn more?

  • Are you working on AI/robotics and need high-fidelity simulations? We’re selecting early partners to fine-tune LoRAs on domain-specific data.
  • Want to explore the future of generative gaming? Sign up for early access to Lucid v2

Let’s connect! Reach us at alberto@lucidsim.co or sign up at lucidsim.co