High Fidelity 3D Reconstruction

Designed and optimized for production.

Easily create photorealistic 3D scenes and objects with any camera in minutes. Faster than traditional NeRF, with richer details and more realistic rendering.

Shoot with any camera

Radiance Field technology creates 3D scenes even from regular RGB images without requiring specialized scanning devices. Reconstruct from photos taken with phones, cameras, or drones.

Fast and easy workflow

Getting started with divshot only takes a few clicks and a set of images. divshot automatically learns the 3D scene from your images.

Full Content Ownership

Your content remains on your devices and under your full control. divshot runs all processing and machine learning steps on devices only you have access to.

Scenes from Photos

๐Ÿ“ธ

Multi-view Reconstruction

Automatically reconstruct 3D scenes from videos or multi-view photos, supporting any camera and shooting angle

๐ŸŽจ

High Fidelity Rendering

High-quality rendering based on Gaussian point clouds, preserving real-world materials and details

โšก

Live Training Preview

View training progress and rendering results in real-time with interactive scene browsing

๐Ÿ’พ

Model Sequence Rendering

Export to PLY, GLTF, OBJ and other common formats, compatible with mainstream 3D software and engines

๐Ÿ”ง

Memory Optimization

Intelligent memory management and LOD hierarchical rendering for large-scale scene reconstruction

๐ŸŽฌ

Unreal Engine Plugin

Seamless integration with UE for real-time virtual production and mixed reality compositing

3DGS Technology

What is 3D Gaussian Splatting?

3DGS (3D Gaussian Splatting) is a revolutionary 3D scene representation method that uses anisotropic Gaussian points to model scenes. Compared to traditional NeRF and MVS methods, 3DGS offers the following advantages:

โšก Faster Rendering

Real-time rendering capability, hundreds of times faster than NeRF

๐ŸŽฏ Richer Details

Preserves high-frequency details and realistic materials for higher reconstruction quality

๐Ÿ’ก More Realistic Visuals

Supports anisotropic reflections and complex lighting effects

๐Ÿ”ง Efficient Storage

Optimized data structures supporting large-scale scenes

Technical Features

  • LOD (Level of Detail) support with dynamic detail adjustment based on viewpoint
  • Intelligent memory management optimizing GPU resource usage
  • Compression algorithms reducing model file size
  • Support for high-resolution input (4K, 8K) and output

Flexible and Scalable

  • Unlimited image count, unlimited model size
  • Supports images of any size, incl. 4K, 8K
  • Immediate local processing
  • Image masking for occluder and background removal
  • Import Poses/Points from 3rd party tools
  • Supports inputs in JPG, PNG, TIF, EXR, DNG, RAW, MOV, MP4, MKV
  • Color space handling, supports ACES
  • Integrates with geo-referenced data
  • Graphical and command-line interfaces
  • Export models to 3rd Party viewers

Ready to Get Started?

Download divshot for free and start creating high-fidelity 3D reconstructions in minutes.

Requires Windows 10+ and NVIDIA RTX GPU. View system requirements โ†’