100,000 OVRMaps 🚀: A Milestone in Decentralized Spatial Computing

This is truly an incredible achievement— +100,000 ultra-detailed 3D Maps (OVRMaps) — a feat that once felt like a distant dream. This milestone is a roaring testament to the power of DePIN (Decentralized Physical Infrastructure Networks) and our global community of mappers. Just one year ago, we were generating only 400 maps monthly. Today, fueled by a revamped incentive system and a tireless technical team, we’re processing 55 TERABYTES of image data with state of the art AI Models to build the foundation of Spatial Computing.

 

Breaking down the Numbers: Why 100k 3D Maps Matter:

Let’s quantify this leap:  

↳ 30 million square meters of real-world terrain mapped in 3D  

↳ 50 million photos captured by our community mappers  

↳ 55 Terabytes of Raw data

↳ 40 GPUs running state-of-the-art AI models to transform raw images into precision maps  

↳ Relevance over randomness: Every OVRMap covers high-traffic, socially vibrant locations—explore them all on the OVER Marketplace.  

 

Adding Perspective:

Currently no Web3 project can match this scale in creating 3D Maps. For a fair comparison we need to switch to Web2 with Niantic – the creator of PokemonGo – as the only player achieving similar magnitude in the creation of 3D Maps with this level of detail. 

And the detail level is paramount, there are multiple companies working on spatial data, not only Niantic but also Google and Mapillary in Web2 or Hive Mapper and Natix in Web3. Yet not all maps are created equal, different data density and geographic coverage enables different use-cases. Spatial data industry splits into two main categories:

 

 

Navigation & Basic VPS:

↳ Use Case: Street-level routing, low-accuracy visual positioning (better than GPS in cities, but limited).  

↳ Data Source: Cars, dashcams, sparse imagery (5–10 photos per 300 sqm).  

↳ Players: Google, Mapillary, HiveMapper.  

 

 

AR, Digital Twins & Precision VPS 

↳ Use Case: Sub-5cm accuracy for AR content anchoring, real-time digital twins, machine perception.  

↳ Data Source: 300–1,000 photos per 300 sqm from pedestrians (OVER) 

↳ Players: OVER, Niantic.  

 

See the Difference between the two categories: ⤵️  

 

For a more complete picture, the following chart summarizes the different features of spatial data generated by some of the most relevant players in the space:

 

An important metric to take into consideration while comparing spatial data is Consistency. Visiting the Niantic’s Lightship Map website and browsing for the activated VPS locations – the number of active active locations is not clear, 1 Mln declared, from our estimates 300k – it is quite apparent that their maps have a very inconsistent dimension and quality, from a single traffic sign to a complete map of an area. This happens because their data have been scraped from different sources, including PokemonGo players without any direct incentive or intention to generate a consistent quality 3D Map. 

OVER’s Edge: first of all, unlike in Web2, mappers are rewarded for the value they provide to the protocol by uploading image data. Moreover, the incentives are not distributed for low importance locations and if the minimum quality requirements for the generated 3DMap are not met. This ensures a consistency level that simply cannot be achieved by Web2 providers who do not distribute any rewards to mappers. 

 

 

What are OVER 3D Maps used for?

OVRMaps database is the foundation of multiple vertical applications in Spatial Computing, let’s explore those one by one:

• Large Geo-Spatial Model (LGM) training: after the incredible success of LLMs we’re assisting the emergence of a new cohort of AI models not only focused in understanding language but in understanding the physical world, they are called Large Geo-Spatial Models. LGMs understand the structure of the 3D world around us just like our brain does, downstream applications goes from re-localization with unstructured images (3DMaps from less data) to spatial segmentation (distinguishing elements inside an image) and coherent 3D Assets generation. Just like for LLMs, LGMs quality heavily depends on the size and quality of the dataset used to train them. OVER can leverage its unique dataset of 3DMaps to generate a state of the art LGM.

• • Geo-localized AR publishing: OVRMaps allows to AR publishers to have a reference of the 3D structure of the locations they are publishing content to. Moreover the visual positioning system allows to AR devices (Smartphones and Smart Glasses) to correctly position content in space both indoors and outdoors, something that is simply not possible with GPS only. Basically 3DMaps are the most critical enabler for geo-localized AR, without 3DMaps only a very limited set of AR applications is possible.
  • Digital Twin creation: thanks to technologies like Gaussian Spating, OVRMaps enable the creation of real time explorable digital twins of physical locations, making it possible to visit physical locations remotely.
  • Merging AR and VR: AR has the incredible potential of augmenting our visual field with coherent 3D assets yet it lacks the “non locality”, the capability of visiting a location remotely typical of VR. OVRMaps combine the potential of the two technical paradigms, making local AR experiences browsable remotely inside a digital twin of the physical location.

• Visual Positioning System (VPS) services: 3D Maps allows for precise re-localization in space, leveraging computer vision algorithms our VPS allows devices to understand where they are in space with Cm accuracy. The applications of this technology is extremely broad going from AR marketing to remote maintenance and ride-sharing fleet re-localization.

• Robotic perception: just like AR devices need to understand where they are in space to coherently augment our visual field, machines need to understand the space around them to successfully operate in it. OVRMaps lay the foundation for machine spatial orientation (VPS) and understanding (LGM).

 

Where do we go from here?

100k 3D Maps is a remarkable objective, yet it is just the beginning, our goal is to become the largest operator worldwide for AR and VPS related services. To achieve this goal we will keep expanding our dataset with continued incentives, including non Web3 natives that will receive rewards in goods and services thanks to our partnership with Bitrefill. Will keep upgrading our mapping capabilities, making the mapping experience faster and smoother. We will leverage LGMs to expand the quality range of data that we can use, allowing for the generation of 3D Maps with more sparse datasets, achieving faster mapping. This will also allow us to leverage open sparse datasets, like Mapillary to expand the coverage of our VPS services, and to partner with operators building complementary datasets, such as Natix. 

 

Join the Movement

To our mappers, developers, and believers: This milestone is yours. Together, we’re not just mapping the world—we’re building the OS of Spatial Computing.  

 

Shape the future of Spatial Computing together

👉 Start Mapping | 👉  Explore OVRMaps