From Far and Near: Perceptual Evaluation of Crowd Representations Across Levels of Detail

2025-10-23

AI Summary

This study addresses the trade-off between visual fidelity and computational efficiency in multi-scale crowd rendering. We systematically investigate users’ subjective perceptual differences across various character representations—geometric meshes, image-based proxies, Neural Radiance Fields (NeRF), and 3D Gaussians—under varying levels of detail (LoD) and observer distances. For the first time, we integrate psychophysical experiments, a structured visual quality questionnaire, and objective performance metrics into a comprehensive, multi-dimensional evaluation framework. Our results identify perceptual fidelity boundaries and computational efficiency breakpoints for each representation at near and far viewing distances. Based on these findings, we propose a perception-driven LoD classification principle that prioritizes human visual sensitivity over geometric or radiometric accuracy. This work provides empirical foundations and actionable optimization guidelines for real-time, visually plausible crowd rendering in interactive applications.

Technology Category

• Humans and AI: Crowd Sourcing and Human Computation
• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

In this paper, we investigate how users perceive the visual quality of crowd character representations at different levels of detail (LoD) and viewing distances. Each representation: geometric meshes, image-based impostors, Neural Radiance Fields (NeRFs), and 3D Gaussians, exhibits distinct trade-offs between visual fidelity and computational performance. Our qualitative and quantitative results provide insights to guide the design of perceptually optimized LoD strategies for crowd rendering.

Problem

• Evaluating visual quality of crowd representations at different LoDs
• Comparing perceptual trade-offs between four rendering techniques
• Providing guidance for perceptually optimized crowd rendering strategies

Innovation

• Evaluating crowd representations across detail levels
• Comparing geometric meshes, impostors, NeRFs, and 3D Gaussians
• Providing perceptual guidance for optimized crowd rendering