The Next Gen Pipeline
Marionette Studios is not just a content creator. We are a startup innovating the production pipeline itself. We connect the disjointed workflows of today's industry into a single, seamless stream driven by AI.
The Problem
Fragmented & Disconnected
Today's production pipeline is broken. Designers in Maya, Engineers in Unreal, and AI tools are all isolated islands. Data conversion costs time and creativity.
Our Mission
One Seamless Stream
We unify the workflow. By connecting the "Eye" (Meta Orion) to the "Brain" (Athena AI), we empower creators to shape reality in real-time, free from technical barriers.
Orion Vision Experience
Real-time rendering streamed directly to AR Glasses via CloudXR.
Core Competitiveness
Domain-Specific AI
An artist-centric workflow designed by a TD with 30 years of experience, not just generic AI.
Hybrid Rendering
Real-time fusion of Rasterization (Unreal/Omniverse) and Generative AI (NIM).
OpenUSD Native
A standardized data flow based on Apple & NVIDIA's OpenUSD.
SYSTEM ARCHITECTURE
Detailed Technical Data Flow & Logic
DCC Tools
Asset Creation
Maya
Unreal 5
Houdini
Nucleus
Data Switch
Live Sync
USD Composer
Asset DB
ATHENA CORE
Middleware
Python Ext
NIMs (AI)
Audio2Face
CloudXR
Streaming
RTX 4090
Low Latency
5G / Wi-Fi 6E
Device
Immersive
Meta Orion
Quest 3
Vision Pro
OpenUSD Standard
Gen AI Processing
Real-time Streaming
Built with Giants
Orchestrating the world's best technologies into one seamless pipeline.
NVIDIA
Computing & Platform
NVIDIA Omniverse & NIM
Intelligence Model
Google Gemini 1.5 Pro
Meta
Extended Reality
Orion Vision & Quest 3
AntiGravity
Core Middleware
Aurora Engine Core
Operational Scenario & Tech Stack
How Aurora Engine connects Reality with AI
.png)
Scene 1: Intro
Streaming high-fidelity UI to lightweight AR glasses via low-latency 5G/Wi-Fi.
TECH: NVIDIA CloudXR + OpenUSD UI
.png)
Scene 2: Persona
Real-time lip-sync generation from audio and intelligent dialogue agent.
TECH: Audio2Face NIM + NVIDIA ACE
.png)
Scene 3: Trigger
Translating vague user commands into precise technical prompts.
TECH: Multimodal Input (Hand Tracking + LLM)
.png)
Scene 4: Magic
Visualizing AI inference time with physical particle simulations.
TECH: Async Processing + Particles
.png)
Scene 5: Result
Generating 3D PBR materials and locking them to real-world coordinates.
TECH: Gen AI (SDXL) + Spatial Anchoring