FacebookFacebook group TwitterTwitter
ICVSS Computer Vision in the Age of Large Language Models

3D Space in Art, Photography, and Human Vision

Aaron Hertzmann

Adobe Research, USA

Abstract

How do we humans perceive 3D space, in pictures and in the world? In the past few decades, human vision science has developed to a very counterintuitive understanding of how human vision interprets the 3D world, which, in part, depends on the operations of foveal and peripheral vision. In pictures, art historical and perceptual theories of perspective have focused on linear perspective, which cannot explain the many different ways that pictures work or how we perceive them. Meanwhile, computational photographic techniques have developed effective algorithms for producing more effective pictures than conventional linear perspective. I'll review all of these developments, and propose a new approach to understanding how humans interpret 3D shape in pictures, and how we make pictures.