Skip to content

NeROIC

Recover object geometry and material from online photos

Visit official site

What neroic is

The method is an object acquisition framework built on a multi-stage extension of neural radiance fields. It infers surface geometry, refines camera parameters and extracts material and lighting from online photo sets.

Main features

  • Inference of surface geometry and camera parameters
  • Robust normal estimation that reduces noise
  • Transient elements such as sharp shadows handled

Pros & cons

✓ What we like

  • Foreground masks improve training efficiency
  • A robust normal estimate keeps key detail

! What to watch out for

  • Input relies on a collection of online images
  • The approach extends neural radiance fields in stages

FAQ

What applications are named?

Novel view synthesis, relighting and harmonious background composition.

How is lighting represented?

With spherical harmonics for environment lighting.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review