Pre-training
Generate millions of physics-ready assets through one API, purpose-built for foundation model training
PRESENTING AT SIGGRAPH 2026 · LOS ANGELES
Build simulation-ready worlds from reality.
Physics is verifiable, the way code is verifiable. Palatial turns real objects and environments into physics-accurate simulation, validated against ground truth and generated at the speed and scale required to train physical AI.
Partners & Customers
01 Applications
Across the physical AI lifecycle.
Reinforcement Learning
Give agents interactive worlds where joints, contact, and failure modes are real.
Evaluations
Run the same task across assets and catch physics, collision, and behavior regressions.
Digital Twinning
Mirror a physical object or cell so sim decisions match what happens on the floor.
02 Headline features
From raw capture to proven simulation.
A coordinated stack of modeling and physics agents turns observations into editable, articulated, verified assets.
03 Asset Library
Simulation-ready objects, ready to drop in.
Inspect a growing catalog of physically accurate assets with clean geometry, collision setup, and exports for the simulators your team already uses.
Preparing 3D preview
D00435 / KEY RACK
Sim-readyKey Rack
A detailed key holder with separate key bundles, metal rings, hooks, and a wooden base.
- Class
- Everyday Objects
- Physics
- 8 rigid parts
- Source
- Palatial Library
- Exports
- USD · MJCF · URDF
04 Supporting features
The details that make assets simulation-ready.
Precision Geometry
Highly detailed shapes with accurate dimensions and collisions
Sim-ready Topology
Feature aligned meshes that work well with physics solvers and decimation
Realistic Materials
High fidelity PBR texture maps and clean UV unwraps for easy editing
Accurate Physics
Mass, inertia, friction, elasticity, with support for soft body and deformables
Interactivity
Articulated objects and end-effectors with joint friction, torque and constraints
Egocentric video to SimReady
Turn real-environment capture into simulation-ready scenes and assets.
06 Testimonials
Trusted by teams shaping what’s next.
From days to hours. From files to spatial understanding. Teams use Palatial to move with confidence.
“Palatial was able to generate a realistic 1:1 representation that enabled us to validate and demonstrate our capability quickly. Generating new assets on the fly unlocks the ability to train models on diverse scenarios we wouldn't be able to do otherwise.”
Jason Ashton
CEO, Aurum Systems
“Palatial has helped us accurately model some very tricky non-rigid objects in simulation. We’re looking forward to continuing to work with them as their platform scales up in the future.”
Vishal Dugar
Co-Founder, Gritt Robotics
07 Team
The team bridging sim to real.
Building at the intersection of robotics, machine learning, graphics, and spatial systems.
Join the team
Founding team / 01
Steven Ren
Founder & CEO
Founding team / 02
Krish Mehta
Founder & CTO
08 Deploy