3D Gaussian Splatting (3DGS) has recently emerged as an efficient representation for real-time novel view synthesis, offering a compelling alternative to traditional neural rendering techniques. By representing scenes as collections of Gaussian primitives, 3DGS enables high-quality rendering with significantly reduced computational cost. Despite these advantages, most existing approaches rely on pixel-level distortion losses, such as L1 or SSIM, during training. These losses are poorly aligned with human visual perception and often result in overly smooth reconstructions that fail to preserve fine textures and structural details. Recent work has demonstrated that replacing such losses with perceptually motivated alternatives can substantially improve visual quality, with human studies showing strong preference for perceptually optimized models over standard approaches. This highlights the importance of designing training objectives that better reflect perceptual quality, rather than relying solely on traditional image fidelity metrics.

Although perceptual losses have shown promising results in improving 3DGS reconstructions, their design and integration remain relatively unexplored. It is still unclear which perceptual distortions are most suitable for 3D Gaussian representations and how these distortions affect both visual quality and model efficiency. In addition, while improvements are often validated through subjective studies, there is limited understanding of how perceptual gains relate to objective metrics and compression efficiency in 3DGS frameworks.

Objectives: The goal of this thesis is to investigate perceptual optimization strategies for 3D Gaussian Splatting and to analyze their impact on rendering quality. The work will focus on studying different classes of distortion losses, including pixel-based, perceptual, and distribution-based losses, and evaluating their effectiveness in preserving fine details and textures. Particular attention will be given to losses inspired by human perception, such as those based on feature statistics or distribution matching. The thesis will also examine how perceptual optimization affects model characteristics, including the number of Gaussians, rendering efficiency, and compression performance. A secondary objective is to analyze the relationship between perceptual improvements and both objective metrics and human preference, using subjective evaluation as a validation tool.

Methodology and Expected Contributions: The research will involve implementing and integrating different loss functions within a 3DGS training pipeline. Experiments will be conducted on standard datasets to compare reconstruction quality across different optimization strategies. The evaluation will combine objective metrics, such as LPIPS and DISTS, with controlled subjective comparisons to assess perceptual quality. Additional analysis will focus on efficiency-related factors, including model size and rendering performance.

This thesis is expected to provide a systematic analysis of perceptual optimization in 3D Gaussian Splatting and to identify effective loss functions for improving visual quality. It will contribute to a better understanding of the trade-offs between perceptual fidelity, computational efficiency, and compression, and provide practical insights for designing perceptually optimized 3D rendering systems.

Contact Saeed.Mahmoudpour@vub.be for more information.

Kerbl, Bernhard, et al. rd gaussian splatting for real-time radiance field rendering." ACM Trans. Graph. 42.4 (2023): 139-1.
Zhang, Richard, et al. "The unreasonable effectiveness of deep features as a perceptual metric." Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
Ghildyal, Abhijay, et al. "Non-Aligned Reference Image Quality Assessment for Novel View Synthesis." Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 2026.
Ozyilkan, Ezgi, et al. "Drop-In Perceptual Optimization for 3D Gaussian Splatting." arXiv preprint arXiv:2603.23297 (2026).

Strong programming skills (preferably Python, and familiarity with PyTorch or similar frameworks)
Basic knowledge of computer vision and/or machine learning
Familiarity with 3D graphics or rendering is a plus, but not strictly required
Interest in experimental research and evaluation (including perceptual assessment)