Skip to content

Latest commit

 

History

History
110 lines (42 loc) · 1.85 KB

Notes.md

File metadata and controls

110 lines (42 loc) · 1.85 KB

Problem

Modeling a real scene from captured images and reproducing its appearance under novel conditions.

recover scene geometry and reflectance

surface reconstruction

Related technology

3D reconstruction & inverse rendering

new break

combines neural scene representations with classical ray marching - a volume rendering approach that is naturally differentiable

  • implicit model
  • explicit model

Literature Contents

[Neural Reflectance Fields](#Neural Reflectance Fields)

[Analytic Meshing](#Analytic Meshing)

Neural Reflectance Fields

Neural Reflectance Fields for Appearance Acquisition

[] (UC San Diego) [SAI BI, ZEXIANG XU] (📝) [:octocat:]

Click to expand

Field

view synthesis works

Noun explanation

reflectance:

light transmittance:

Aims

https://whimsical.com/view-synthesis-2zfiLatcYU3nsNmAb2y74f

Analytic Meshing

Analytic Marching: An Analytic Meshing Solution from Deep Implicit Surface Networks

[PMLR 2020] (SUST) [Jiabao Lei, Kui Jia] (📝) [:octocat:]

Click to expand

Summary

This paper studies the problem of surface reconstruction through learning surface mesh via implicit functions, where implicit functions are implemented as multi-layer perceptrons (MLPs) with rectified linear units (ReLU).

Related work

  • marching cubes (standard algorithm)
  • a ReLU based MLP partitions its input space into a number of linear regions
  • polygonal mesh
  • signed distance function or SDF

isosurface extraction: implicit representation -> explicit surface mesh