Photography Perspective Composition: Towards Aesthetic Perspective Recommendation

TL;DR

This paper introduces a novel perspective adjustment method for photography composition that enhances aesthetic balance by modifying 3D perspectives, supported by datasets, transformation videos, and a quality assessment model.

Contribution

It presents a comprehensive framework for perspective composition, including dataset creation, transformation visualization, and a perspective quality assessment model, advancing beyond traditional cropping methods.

Findings

Automated dataset construction from expert photographs

Video demonstrations of perspective transformation

A human performance-based perspective quality assessment model

Abstract

Traditional photography composition approaches are dominated by 2D cropping-based methods. However, these methods fall short when scenes contain poorly arranged subjects. Professional photographers often employ perspective adjustment as a form of 3D recomposition, modifying the projected 2D relationships between subjects while maintaining their actual spatial positions to achieve better compositional balance. Inspired by this artistic practice, we propose photography perspective composition (PPC), extending beyond traditional cropping-based methods. However, implementing the PPC faces significant challenges: the scarcity of perspective transformation datasets and undefined assessment criteria for perspective quality. To address these challenges, we present three key contributions: (1) An automated framework for building PPC datasets through expert photographs. (2) A video generation approach that demonstrates the transformation process from less favorable to aesthetically enhanced perspectives. (3) A perspective quality assessment (PQA) model constructed based on human performance. Our approach is concise and requires no additional prompt instructions or camera trajectories, helping and guiding ordinary users to enhance their composition skills.