XReco MVP: Democratizing 3D Model Reconstruction and Beyond

As XReco enters its next testing phase, we’re not just validating features — we’re bringing an innovative platform to life, pushing the boundaries of 3D reconstruction and beyond! 

The new release of the XReco platform is coming! MVP stands for Minimum Viable Product, and, in the context of a research project, refers to a prototype designed to test the feasibility of an innovative concept with minimal development effort. 

In XReco, we have structured the presentation and validation process into two distinct phases. Around the midpoint of the project, we conducted the first validation phase, where we gathered a large group of friendly users to test partially integrated platform services, as well as some individual components in isolation.

Now, in mid-April, we are preparing to launch the second iteration of this process. This time, our main objective is to ensure that all platform services are integrated in some way and presented to users in a consolidated manner, allowing us to evaluate their combined performance. The final stage of this validation round will involve in-person, unguided testing, where users will have the opportunity to freely explore XReco’s feasibility, providing insights into its real-world usability and effectiveness, having some kind of a market validation feedback.

By developing an MVP, research teams can refine methodologies, evaluate performance under real-world conditions, and lay the groundwork for further advancements. This process is essential in ensuring that research outputs transition effectively from theoretical exploration to practical implementation.

The deployment of the XReco platform has been a collaborative integration effort, with all technology partners contributing with their expertise. The platform consists of a distributed deployment, where each partner has provided part of their infrastructure to deliver a high-performance service capable of meeting intensive computational demands while remaining accessible to users.

Figure 1: Current Deployment of the XReco MVP

But, what can users expect from the platform? XReco is conceived as a way to democratize access to state-of-the-art 3D model reconstruction from 2D media sources. However, its capabilities go far beyond that, evolving into a multifunctional platform that integrates a wide range of features, including:

• Automatic landmark and object detection: Identify landmarks and objects instantly with AI-powered recognition.
• Automated tagging for journalistic articles: Enhance journalism search with AI-driven article tagging for better organization and discovery.
• Multimodal search: Find related 3D models, videos, and images using any media file as input.
• Secure user management and licensing options:  Ensure safe access and content protection with robust user and licensing controls.
• A blockchain-based marketplace: Buy, sell, and trade digital assets securely with blockchain-powered transactions.
• Integration with multiple external content sources: The only openly available metasearch engine unifying 3D models, images, and videos in a single result.

Figure 2: XReco orchestrator dashboard, where several multimodal results and reconstruction services are available

3D reconstruction services are transforming multiple market verticals, enabling new possibilities in industries ranging from media production to digital heritage preservation. At the core of this evolution, XReco focuses on state-of-the-art reconstruction services, trained and customized for seamless activation with just a few clicks. These include:

• Structure from Motion: Reconstruct 3D structures from 2D images using motion analysis.
• NeRF in the Wild: Generate realistic 3D scenes from unstructured real-world photos.
• Instant-NGP: Fast and efficient neural graphics primitives for real-time 3D rendering.
• 3D GS: Radiance field rendering from 2D inputs.
• GDNeRF + Depth: Efficient view synthesis for humans using depth-based NeRFs 
• GDGS: Human-based:  Human-centric generative modeling for realistic 3D avatars and environments based on Gaussian Splatting  

The reconstructed 3D models can be published on the XReco marketplace, shared with other users, fostering a dynamic ecosystem for creating and exchanging 3D experiences. But the user journey doesn’t stop there. XReco is also integrated with two XR-based authoring tools, designed to bridge the gap between 3D content creation and immersive experiences:

• XR Capsules: A tool that serves as a launchpad for users with little experience in 3D, making it easier to create XR scenes compatible with game engines like Unity.
• XR CMS: Enables users to create immersive experiences by geolocating assets extracted from XReco, placing them in specific points within extended reality environments.

With this level of integration, XReco not only facilitates 3D model reconstruction but also empowers users to build, share, and experience next-generation digital content.

Sign up for exclusive MVP testings in May 2025.

XReco is entering its second validation phase in mid-April, to demonstrate the feasibility, usability, and market readiness of its platform. By integrating state-of-the-art reconstruction services and blockchain-based marketplaces, XReco aims to democratize access to advanced 3D model creation while offering secure licensing and user management. The platform’s distributed deployment unites multiple partners’ infrastructures and seamlessly connects with XR authoring tools for extended-reality experiences. This iterative process exemplifies how research-driven MVPs can transition from innovative proof-of-concept to practical, user-friendly solutions in real-world scenarios.

Ries, E. (2011). The Lean Startup: How Today’s Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses. Crown Business.

Antonio Calvo: Antonio is a Spanish R&D engineer who has worked on television infrastructure deployment for many years. He is currently part of the media-tech department at i2Cat, providing holographic solutions and reconstruction services. His main role in the XReco project has been coordinating the global integration and specifically the development of the platform’s metasearch engine.

Never miss updates on the latest XReco developments.