[arXiv]score: 0.38
Aes3D: First Framework for Aesthetic Quality Assessment of 3DGS Scenes
May 26, 2026
Aes3D introduces the first dataset and framework for evaluating aesthetic attributes (composition, harmony, visual appeal) in 3D Gaussian Splatting scenes, addressing a gap left by reconstruction-fidelity-focused metrics.
cs.CVcs.AI
HOW THIS AFFECTS YOU
●
researcherProvides the first annotated aesthetic dataset for 3DGS, enabling new benchmarks beyond PSNR/SSIM-style fidelity metrics.
●
designerYou can potentially use this framework to evaluate and improve visual quality of 3D scenes beyond technical reconstruction accuracy.