Notifications
Back

RECON Labs' technology represents objects and spaces in an innovative way.

Discover the seamless bridging between the real and the virtual world.

Comapany Overview

Core Tech: NeRF

(Neural Radiance Fields)

RECON Labs offers solutions based on NeRF (Neural Radiance Fields) to restore 3D spatial information in a revolutionary way. NeRF utilizes a neural network for the reconstruction of 3D shapes using only typical camera images, without the need for additional distance measuring devices. This method allows for highly accurate restoration of 3D shapes or navigation of 3D spaces.

Comapany Overview

Principle

Each pixel that composes the image contains color information of the light entering the camera sensor in space. NeRF is a technique for inferring spatial information through a neural network based on the color of a large number of pixels in multiple images. A neural network applied with the latest computer vision and graphics AI technology infers optical information such as color or density of each point in space, and learns by comparing how similar the image created based on the inferred spatial information is to the actual filmed video. RECONLABS provides various solutions using the AI-based 3D information expression method built in this way.

RECON Tech

MetaRECON:

3D Reconstruction 

AI Technology

Traditional photogrammetry techniques often struggle to reconstruct 3D shapes of objects with textureless or reflective surfaces.

To address this challenge, RECON Labs has developed MetaRECON, a 3D reconstruction engine that combines the traditional photogrammetry pipeline with AI-based technologies to improve surface quality. As a result, MetaRECON is able to accurately reconstruct even the most complex surfaces in 3D.

RECON Tech

Seamless Web-Based 3D Content Visualization

RECON Labs develops a technology that reverse-engineers video footage to make it appear 3D, surpassing traditional polygon-based model visualization methods with a lightweight, high-quality viewer. This technology allows for real-time visualization on the web, compatible with Android, iOS, and desktop platforms, running at a smooth 30-60 frames per second. Our approach relies on lightweight models (10-20MB) and supports web-based scene editing, including cutting and color adjustments.


We implement VFX visual effects on a point-by-point basis, enhancing realism. With these technological advancements, we build an environment for swiftly generating, visualizing, and editing high-quality 3D content on the web.

RECON Labs technology represents objects and spaces in an innovative way.


Discover the seamless bridging between the real and the virtual world.


Core Tech: NeRF

(Neural Radiance Fields) 

RECON Labs offers solutions based on NeRF (Neural Radiance Fields) to restore 3D spatial information in a revolutionary way. NeRF utilizes a neural network for the reconstruction of 3D shapes using only typical camera images, without the need for additional distance measuring devices. This method allows for highly accurate restoration of 3D shapes or navigation of 3D spaces.

Principle

Each pixel that composes the image contains color information of the light entering the camera sensor in space. NeRF is a technique for inferring spatial information through a neural network based on the color of a large number of pixels in multiple images. A neural network applied with the latest computer vision and graphics AI technology infers optical information such as color or density of each point in space, and learns by comparing how similar the image created based on the inferred spatial information is to the actual filmed video. RECONLABS provides various solutions using the AI-based 3D information expression method built in this way.

RECON Tech

MetaRECON: 

3D Reconstruction AI Technology

Traditional photogrammetry techniques often struggle to reconstruct 3D shapes of objects with textureless or reflective surfaces.
To overcome the difficulty in reconstructing 3D shapes of objects with textureless or reflective surfaces, RECON Labs has developed MetaRECON.
This 3D reconstruction engine combines traditional photogrammetry with AI-based technologies to produce high-quality surface reconstructions.
Using Agisoft Metashape's cloud-based camera pose calculation process, we efficiently align images. MetaRECON is designed for training the volume field through NeRF and also enhances the detail and stability of the resulting 3D mesh by refining the NeRF training results into an SDF function.
As a result, MetaRECON can create more accurate and stable reconstructions than other solutions, even for the most complex surfaces.

RECON Tech

Seamless Web-Based 3D Content Visualization

RECON Labs develops a technology that reverse-engineers video footage to make it appear 3D, surpassing traditional polygon-based model visualization methods with a lightweight, high-quality viewer. This technology allows for real-time visualization on the web, compatible with Android, iOS, and desktop platforms, running at a smooth 30-60 frames per second.

Our approach relies on lightweight models (10-20MB) and supports web-based scene editing, including cutting and color adjustments.

We implement VFX visual effects on a point-by-point basis, enhancing realism. With these technological advancements, we build an environment for swiftly generating, visualizing, and editing high-quality 3D content on the web.