Home / MORPHEUS
Private Sector
Stage 1, Stage 2

MORPHEUS

Phasmatic Private Company
‹ Back to Open Call Winners
MORPHEUS turns a short video or 30–150 photos of a physical object into a textured, engine-ready 3D model (glTF/FBX) in a few minutes. It uses self-supervised neural surface reconstruction and needs no training data. Already used in three EU projects (200+ objects, 20+ users), it serves cultural heritage, XR, education, robotics and 3D e-commerce. In DeployAI, Phasmatic lists MORPHEUS on the AIoD Marketplace with metered, tiered access, runs high-resolution jobs on MeluXina via the GPU Starter Plan, checks its minimal-risk AI Act class with TAI Scan, and prepares a Stage 3 pilot with the M-Cube Foundation. All processing stays in the EU.

Background

Getting a physical object into a 3D application is still slow, costly and specialist work. A 3D artist can spend hours modelling a single object by hand. Professional photogrammetry suites and scanners are expensive and still need manual clean-up. Research neural-reconstruction tools produce impressive renders, but not a clean mesh that a game engine, AR viewer or online shop can use. Museums, educators, XR developers, robotics teams and online retailers all hit this bottleneck: they need many ready-to-use 3D models of real objects, quickly, affordably and with data kept in the EU.

Proposed solution

MORPHEUS is a cloud-native 3D reconstruction service. Users upload a 20-60s video or 30-150 photos of a small object, taken with a phone or camera. The pipeline runs in four stages inside one container: image calibration (frame selection, background removal, camera pose estimation); neural surface extraction, where a signed distance field is fitted to the images by differentiable volume rendering; asset optimisation (mesh clean-up, simplification, UV unwrapping, texture baking); and export. The result is a textured triangle mesh in glTF or FBX that opens directly in Unity, Unreal, Blender or a web viewer. A standard job takes about 7 minutes on one 16 GB GPU; high-resolution jobs run on EuroHPC capacity. The same Docker image runs as SaaS, on a customer’s own servers or on HPC. A content-safety pre-filter blocks images of identifiable people, inputs are deleted after processing and customer data is never used for training. Access is pay-as-you-go, by subscription, free for academic use, or via a self-hosted enterprise licence.

Objectives

  • Turn a short video or photos of a physical object into a textured, engine-ready 3D mesh in a few minutes
  • Fill a gap in the AIoD Marketplace with a ready-to-use service for turning physical objects into 3D assets
  • Run on European infrastructure: EuroHPC supercomputing, EU clouds or on-premise, so data stays in Europe
  • Stay trustworthy: minimal risk under the EU AI Act, EU-only processing, customer data never used for training
  • Serve XR, cultural heritage, education and e-commerce teams across Europe

Phasmatic Private Company

Greece
Phasmatic specializes in advanced 3D visualization and digital experience creation across sectors including e-commerce, education, and cultural heritage. We design high-fidelity 3D digital representations and photorealistic, interactive visualizations that enable users to explore, understand, and engage with objects, environments, and narratives in meaningful ways.
Stay informed on AI opportunities
Get updates on new solutions, market trends, and adoption strategies delivered to your inbox.
The DeployAI project has received funding from the Digital Europe Programme (DIGITAL) under grant agreement №. 101146490 - DIGITAL-2022-CLOUD-AI-B-03
Stay informed on AI opportunities
Get updates on new solutions, market trends, and adoption strategies delivered to your inbox.
The DeployAI project has received funding from the Digital Europe Programme (DIGITAL) under grant agreement №. 101146490 - DIGITAL-2022-CLOUD-AI-B-03