Assessing Photorealism of Rendered Objects in Real-World Images: A Transparent and Reproducible User Study

TL;DR

This study evaluates the photorealism of virtual objects integrated into real images using a transparent user study, revealing challenges in distinguishing rendered objects from real ones and setting a foundation for future improvements.

Contribution

It introduces a reproducible user study methodology to assess photorealism of rendered objects in real images, emphasizing transparency and validation.

Findings

Observers struggled to differentiate between real and rendered objects.

The study validates the methodology with a control group.

Results highlight the need for improved rendering techniques.

Abstract

In an era where numerous studies claim to achieve almost photorealism with real-time automated environment capture, there is a need for assessments and reproducibility in this domain. This paper presents a transparent and reproducible user study aimed at evaluating the photorealism of real-world images composed with virtual rendered objects, that have been generated using classical environment capturing and rendering techniques. We adopted a two-alternative forced choice methodology to compare pairs of images created by integrating virtual objects into real photographs, following a classic pipeline. A control group with defined directional light parameters was included to validate the study's correctness. The findings revealed some insights, suggesting that observers experienced difficulties in differentiating between rendered and real objects. This work establishes the groundwork for future studies, aimed at enhancing the visual fidelity and realism of virtual objects in real-world environments.