Pith. sign in

REVIEW 5 major objections 6 minor 52 references

3DGS-IEval-15K: A Large-scale Image Quality Evaluation Database for 3D Gaussian-Splatting

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read 3DGS-IEval-15K provides 15,200 human-scored images for compressed 3D Gaussian Splatting quality.

desk verdict The largest 3DGS IQA dataset to date, useful if the MOS labels hold up; three fixable issues (arithmetic slip, missing reliability stats, over-read view-dependence) stand between this and a default benchmark. read the letter →

arxiv 2506.14642 v2 pith:AXZOZ5W7 submitted 2025-06-17 cs.CV

Yuke Xing , Jiarui Wang , Peizhi Niu , Wenjie Huang , Guangtao Zhai , Yiling Xu This is my paper · ORCID
classification cs.CV
keywords Imagequalityassessment3DGaussianSplatting3DGScompressionSubjectiveevaluationMeanopinionscoreView-dependentIQAbenchmarkNovelviewsynthesis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper introduces 3DGS-IEval-15K, a dataset of 15,200 rendered images that pairs compressed 3D Gaussian Splatting scenes with human quality scores. It is the first image-quality-assessment resource built specifically for compressed 3DGS representations, and it is large enough to train specialized quality metrics rather than reusing generic ones. The authors also benchmark 30 existing quality metrics on the dataset, showing that learned metrics track human scores better than handcrafted ones and that all models lose accuracy when distortions are isolated by type. The dataset matters because 3DGS storage is a bottleneck for deployment, and without a perceptual ground truth, compression choices cannot be tuned to what viewers actually see.

What carries the argument

The central object is the dataset itself, built on a three-part construction pipeline. Distortion design is the load-bearing mechanism: for the four multi-parameter algorithms, geometry and color compression levels are varied on four-point scales and combined pairwise into 16 distortion levels, while the two single-parameter algorithms receive 6 levels each, producing systematic coverage of 3DGS-specific artifacts. Viewpoint selection uses k-means clustering over position and direction for training views and a four-criterion composite score for test views, isolating view-dependent quality. Human scores are collected with a double-stimulus impairment scale and processed with ITU-R BT.500 outlier rejection to yield MOS values; the benchmark then measures 30 metrics with SRCC, PLCC, and KRCC. The construction is what allows the paper to attribute quality differences to distortion type and viewpoint rather than to scene content.

What would settle it

Re-run the subjective experiment on a random subset of about 500 images with a different group of at least 30 viewers; if the two labs' MOS rankings agree with SRCC below 0.9, the single-lab labels are too unstable to support the benchmark's metric rankings.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that human perception of compressed 3DGS can be measured systematically, and that doing so reveals patterns generic IQA metrics miss. The authors construct 720 trained 3DGS models from 10 real scenes using 6 compression algorithms, render each from 20 viewpoints chosen to include both representative and challenging novel views, and collect 15 ratings per image from 60 viewers to produce MOS scores. They find that novel viewpoints receive systematically lower quality scores than training viewpoints, and that while the best deep-learning IQA metrics reach SRCC above 0.93 on the full dataset, every method degrades on geometry-only or color-only distortion subsets. From this they argue that the dataset provides the scale and distortion coverage needed to develop 3DGS-specific perceptual metrics.

Load-bearing premise

The dataset's value rests on the 60 student viewers' ratings being trustworthy ground truth, but the ratings come from one lab with no reported inter-subject agreement, confidence intervals, or second-lab validation.

Editorial extensions

If this is right

  • Specialized 3DGS quality metrics can now be trained on 15,200 labeled images instead of relying on generic IQA models.
  • View-dependent quality is a measurable property of 3DGS: novel viewpoints score lower than training viewpoints, so viewpoint sampling is itself a quality factor.
  • Existing metrics are not interchangeable on 3DGS content: deep-learning models lead the benchmark while BRISQUE and several LLM zero-shot models correlate weakly with human scores.
  • Geometry-only and color-only distortion subsets expose generalization gaps, meaning metric evaluations on mixed distortions can hide weaknesses.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural next step the authors do not take is training a quality regressor on the Gaussian attributes (positions, opacities, SH coefficients) rather than on rendered pixels; the dataset's compression-level design makes such attribute-level supervision possible.
  • The paired geometry and color distortion levels could support ablation studies separating which compression parameter most affects perceived quality, and possibly guide bit allocation between geometry and color during compression.
  • If the view-dependent MOS disparity replicates across more scenes, viewpoint selection in 3DGS training could be adapted to spend more capacity on directions that viewers find perceptually vulnerable.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper introduces 3DGS-IEval-15K, a large-scale image quality assessment (IQA) dataset for compressed 3D Gaussian Splatting. It describes the selection of 10 real scenes from Mip-NeRF 360, Tanks & Temples, and Deep Blending; a viewpoint selection strategy using k-means clustering for training viewpoints and a four-criterion composite score for testing viewpoints; the generation of distortions by training six 3DGS algorithms at multiple compression levels; a subjective experiment with 60 viewers and 15 ratings per image using an 11-level impairment scale; and a benchmark of 30 IQA metrics. The authors claim this is the first large-scale 3DGS-specific IQA dataset, containing 15,200 images, and use it to analyze view-dependent quality differences between training and novel viewpoints.

Significance. If the dataset is valid, it is a substantial community resource. It is considerably larger than prior NVS-QA datasets (e.g., GSC-QA with 120 samples and ENeRF-QA with 440), covers six compression algorithms with separate geometry and color distortion levels, and the authors publicly release the data. The 30-metric benchmark, including deep-learning and LLM-based methods, provides a useful baseline for future 3DGS quality assessment research, and the view-dependent MOS analysis, if confirmed with proper statistical controls, could inform 3DGS optimization. However, the benchmark rankings and the view-dependence conclusions inherit the quality of the subjective labels, and the current manuscript does not yet establish that quality. The reproducible public release is a clear strength, but the arithmetic error in the model count and the missing MOS reliability analysis need to be addressed before the resource can serve as a trusted benchmark.

major comments (5)
  1. [3.3.2] The model-count arithmetic is inconsistent. The paper states 'we train 10 scenes×(4×16+2×6) DLs = 720 3DGS models', but 4×16+2×6 = 76, and 10×76 = 760, not 720. Since 760×20 = 15,200, the correct model count is 760; the printed '720' would imply 14,400 images. This error affects the headline dataset size and must be corrected wherever the count appears.
  2. [3.4] MOS reliability is not established. The manuscript reports 60 student viewers, 15 ratings per image, an 11-level impairment scale, and a 2% ITU-R BT.500 rejection rate, but provides no inter-subject agreement (e.g., ICC, Krippendorff's alpha), per-image standard deviations, confidence intervals, or validation against a second lab. With only 15 naive ratings per image, label noise could change the Table 4 metric rankings and the Figure 4(b) view-dependence comparison. Please report agreement statistics and the exact screening procedure, and justify the unusually low 2% rejection rate.
  3. [4.3.1] The four dataset configurations are not defined precisely enough to rule out leakage. The 'All' configuration is formed by concatenating the train and test sets of Geometry-Only, Color-Only, and Geometry & Color Mix, but the paper does not state whether these subsets are disjoint image sets, nor whether the 4:1 split is performed at the scene level or the image level. If the same source scene appears in both a model's training and test sets, the reported SRCC/PLCC values can be inflated. Please specify the split granularity and explicitly confirm that the three distortion subsets are disjoint.
  4. [Table 4] For InternVL2.5 (8B) and Gemini1.5-pro, the reported SRCC values (≈0.67–0.71) and PLCC values (≈0.08–0.10) are implausibly far apart. Such a large discrepancy is unexpected for a quality prediction task and suggests a nonlinearity or an evaluation artifact in the reference-score alignment. The paper should explain the alignment/calibration procedure for LLM outputs, or these entries should be re-evaluated and corrected.
  5. [4.2 / Figure 4(b)] The claim that test viewpoints yield lower MOS than training viewpoints is based on two overlaid histograms without a significance test or per-scene control. Since viewpoint selection and scene content are confounded, a paired or mixed-effects analysis stratified by scene is needed before describing this as 'the first systematic investigation' of view-dependent quality in 3DGS. Please provide per-scene statistics and a formal test (e.g., a scene-blocked comparison).
minor comments (6)
  1. [3.4] The standards are mislabeled: 'ITUR BT.500' should be 'ITU-R BT.500' and 'ITU-TP.910' should be 'ITU-T P.910'.
  2. [3.2] The symbol S_j is used both for the composite score in Eq. (1) and for the sparsity criterion introduced immediately after; using distinct symbols for the composite score and the sparsity term would remove ambiguity.
  3. [Figure 1(d)] Figure 1(d) states '15 Annotators per image', while the text says 15,200 images are randomly divided into 8 groups and rated by 60 students; please clarify how the 228,000 annotations are distributed across viewers and sessions, and whether each image is rated exactly 15 times.
  4. [4.3.1] The definitions of 'Geometry-Only', 'Color-Only', and 'Geometry & Color Mix' should state explicitly which compression levels are held fixed for the other distortion type, since the multi-parameter algorithms combine 16 distortion levels through pairwise combinations.
  5. [Table 4] No confidence intervals or significance tests are reported for the SRCC/PLCC/KRCC values; at least bootstrap intervals for the top-performing metrics would strengthen the benchmark comparisons.
  6. [3.3.1] The paper alternates between 'Scaffold' and 'Scaffold-GS', and between 'LightGS' and 'LightGaussian'; please unify the naming throughout the text and tables.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the dataset construction, MOS collection, and external-metric benchmark are independent empirical steps.

full rationale

The paper's load-bearing content is a new IQA dataset with human MOS labels and a benchmark of 30 existing IQA metrics. The MOS values are obtained from a controlled subjective experiment (Section 3.4), and the benchmark evaluates external metrics on held-out configurations (Section 4.3); no equation fits a parameter to the target claim and then re-predicts it. The viewpoint selection criteria in Eqs. (1-5) are geometric definitions of 'challenging' viewpoints, and the observation that test viewpoints have lower MOS is an empirical outcome of the measured human ratings, not a quantity defined by the selection rule. The 'first large-scale' claim is supported by a comparative table of sample counts; the self-citations (ENeRF-QA [39], GSC-QA [41]) appear only as related-work context with much smaller datasets and are not load-bearing for any derivation. Remaining concerns, such as the absence of inter-subject agreement statistics for the MOS and the internal arithmetic inconsistency between '720 3DGS models' and the 15,200 images / 760 trained models stated elsewhere, are data-quality and correctness issues rather than circularity. Overall, the central derivation is self-contained and externally benchmarked.

Assumptions & free parameters 5 free parameters · 8 assumptions · 0 invented entities

The central claim is an empirical resource, so the ledger records design choices and unverified assumptions rather than derivation constants. The distortion space is defined by hand-set compression levels, the viewpoint selection uses several hand-set weights, and the validity of MOS as ground truth is assumed. None of these choices is fitted to the target result, so the circularity burden is low; however, sensitivity to the choices is not reported.

free parameters (5)
  • Training viewpoint feature balance beta = 0.3
    Chosen by hand in Section 3.2 to weight viewing direction against camera position; no sensitivity analysis is reported.
  • Composite viewpoint score weights w_d, w_s, w_e, w_theta = 0.25 each
    Set equal by hand in Equation (1); no ablation or justification is given.
  • Neighborhood size K and epsilon = K = min(10, |T|), epsilon small
    Constants used in the sparsity term (Equation 3); no sensitivity analysis is reported.
  • Compression level parameter tables = Table 2 (4x4 CLs) and Table 3 (6 CLs)
    Hand-selected parameter values for the six 3DGS algorithms; these define the distortion space but are not fitted to MOS.
  • Viewpoints per scene = 20 (10 training, 10 testing)
    Design choice for dataset scale; no evidence that 20 viewpoints per scene capture scene-level quality variation.
assumptions (8)
  • domain assumption Averaging ITU-R BT.500 filtered ratings into a per-image MOS yields interval-scale ground truth.
    Section 3.4 applies BT.500 rejection and ITU-T P.910 11-level impairment scale; no check of whether the resulting MOS meets interval-scale assumptions or is stable across viewer subgroups.
  • domain assumption Fifteen ratings per image from 60 students are sufficient for reliable MOS.
    Section 3.4 and Figure 1(d); no per-image variance, inter-class correlation, or subject-rejection statistics beyond a 2% rate are reported.
  • domain assumption Static side-by-side still images capture the perceived quality of interactive 3DGS renderings.
    Section 3.4 uses a double-stimulus still-image protocol although 3DGS is a real-time interactive rendering technology; temporal and interactive artifacts are not assessed.
  • domain assumption The 10 selected scenes from Mip-NeRF 360, Tanks and Temples, and Deep Blending represent the range of real-world 3DGS deployment.
    Section 3.1 selects 10 scenes; diversity is argued with spatial information and colorfulness only, with no quantitative evidence that this set covers the 3DGS use space.
  • domain assumption The viewpoint selection criteria (Equations 1 to 5) identify representative and challenging views in a way that supports quality evaluation.
    Section 3.2 introduces hand-set weights and heuristics; there is no ablation or validation that these criteria correlate with reconstruction difficulty rather than with the chosen metrics.
  • domain assumption Training each algorithm at the Table 2 and Table 3 compression levels produces the intended monotonic distortion levels.
    Section 3.3 reports no bitrate, model size, or reconstruction error for the trained models, so the actual quality ordering of the 16 and 6 distortion levels is not verified.
  • domain assumption Off-the-shelf implementations of the six compression algorithms behave as described in their original papers.
    Section 3.3 relies on public implementations; no code versions, hyperparameters, or training budgets are listed.
  • standard math Equations 2 to 5 are standard geometric and rank-based constructs whose linear combination is valid.
    Section 3.2 uses normalized distances, k-means, convex hull membership, and arccos direction novelty; these are standard, but the composite weighting is an unvalidated modeling choice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 3DGS-IEval-15K: A Large-scale Image Quality Evaluation Database for 3D Gaussian-Splatting." pith.science (2026). https://pith.science/paper/AXZOZ5W7

@misc{pith2026250614642,
  author       = {Pith},
  title        = {Pith review of: 3DGS-IEval-15K: A Large-scale Image Quality Evaluation Database for 3D Gaussian-Splatting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AXZOZ5W7}},
  note         = {Machine review of arXiv:2506.14642}
}
read the original abstract

3D Gaussian Splatting (3DGS) has emerged as a promising approach for novel view synthesis, offering real-time rendering with high visual fidelity. However, its substantial storage requirements present significant challenges for practical applications. While recent state-of-the-art (SOTA) 3DGS methods increasingly incorporate dedicated compression modules, there is a lack of a comprehensive framework to evaluate their perceptual impact. Therefore we present 3DGS-IEval-15K, the first large-scale image quality assessment (IQA) dataset specifically designed for compressed 3DGS representations. Our dataset encompasses 15,200 images rendered from 10 real-world scenes through 6 representative 3DGS algorithms at 20 strategically selected viewpoints, with different compression levels leading to various distortion effects. Through controlled subjective experiments, we collect human perception data from 60 viewers. We validate dataset quality through scene diversity and MOS distribution analysis, and establish a comprehensive benchmark with 30 representative IQA metrics covering diverse types. As the largest-scale 3DGS quality assessment dataset to date, our work provides a foundation for developing 3DGS specialized IQA metrics, and offers essential data for investigating view-dependent quality distribution patterns unique to 3DGS. The database is publicly available at https://github.com/YukeXing/3DGS-IEval-15K.

Figures

Figures reproduced from arXiv: 2506.14642 by the authors.

Figure 1
Figure 1. Overview of the construction pipeline of 3DGS-IEval-15K, designed for benchmarking 3DGS representations. (a) [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. 10 selected source content in 3DGS-IEval-15K: scenes 1-5 depict outdoor scenes, while scenes 6-10 depict indoor scenes. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the proposed viewpoints selection strategy: Training viewpoints are selected through feature-based [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: MOS distribution for both training and testing view [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 27 canonical work pages

  1. [1]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, Humen Zhong, Yuanzhi Zhu, Mingkun Yang, Zhaohai Li, Jianqiang Wan, Pengfei Wang, Wei Ding, Zheren Fu, Yiheng Xu, Jiabo Ye, Xi Zhang, Tianbao Xie, Zesen Cheng, Hang Zhang, Zhibo Yang, Haiyang Xu, and Junyang Lin. 2025. Qwen2.5-VL Technical Rep...

    arXiv 2025

  2. [2]

    Jonathan T Barron, Ben Mildenhall, Dor Verbin, Pratul P Srinivasan, and Peter Hedman. 2022. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR). 5470–5479

    work page 2022

  3. [3]

    ITU Recommendation BT. 2019. 500-14, Methodologies for the subjective assess- ment of the quality of television images.Geneva: International Telecommunication Union (2019)

    work page 2019

  4. [4]

    Yihang Chen, Qianyi Wu, Weiyao Lin, Mehrtash Harandi, and Jianfei Cai. 2024. Hac: Hash-grid assisted context for 3d gaussian splatting compression. In Euro- pean Conference on Computer Vision (ECCV) . Springer, 422–438

    work page 2024

  5. [5]

    Zhe Chen, Weiyun Wang, Yue Cao, Yangzhou Liu, Zhangwei Gao, Erfei Cui, Jinguo Zhu, Shenglong Ye, Hao Tian, Zhaoyang Liu, Lixin Gu, Xuehui Wang, Qingyun Li, Yimin Ren, Zixuan Chen, Jiapeng Luo, Jiahao Wang, Tan Jiang, Bo Wang, Conghui He, Botian Shi, Xingcheng Zhang, Han Lv, Yi Wang, Wenqi Shao, Pei Chu, Zhongying Tu, Tong He, Zhiyong Wu, Huipeng Deng, Jia...

    arXiv 2025

  6. [6]

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P Simoncelli. 2020. Image quality assessment: Unifying structure and texture similarity.IEEE transactions on pattern analysis and machine intelligence 44, 5 (2020), 2567–2581

    work page 2020

  7. [7]

    Zhiwen Fan, Kevin Wang, Kairun Wen, Zehao Zhu, Dejia Xu, Zhangyang Wang, et al. 2024. Lightgaussian: Unbounded 3d gaussian compression with 15x reduc- tion and 200+ fps. Advances in neural information processing systems 37 (2024), 140138–140158

    work page 2024

  8. [8]

    Sharath Girish, Kamal Gupta, and Abhinav Shrivastava. 2024. Eagles: Efficient accelerated 3d gaussians with lightweight encodings. In European Conference on Computer Vision. Springer, 54–71

    work page 2024

Show all 52 references
  1. [9]

    Alireza Golestaneh, Saba Dadsetan, and Kris M

    S. Alireza Golestaneh, Saba Dadsetan, and Kris M. Kitani. 2022. No-Reference Im- age Quality Assessment via Transformers, Relative Ranking, and Self-Consistency. arXiv:2108.06858 [eess.IV] https://arxiv.org/abs/2108.06858

    2022 arXiv

  2. [10]

    Google Team. 2024. Gemini1.5-Pro. https://gemini.google.com/

    2024

  3. [11]

    Peter Hedman, Julien Philip, True Price, Jan-Michael Frahm, George Drettakis, and Gabriel Brostow. 2018. Deep blending for free-viewpoint image-based ren- dering. ACM Transactions on Graphics (ToG) 37, 6 (2018), 1–15

    2018

  4. [12]

    Wenyi Hong, Weihan Wang, Qingsong Lv, Jiazheng Xu, Wenmeng Yu, Junhui Ji, Yan Wang, Zihan Wang, Yuxuan Zhang, Juanzi Li, Bin Xu, Yuxiao Dong, Ming Ding, and Jie Tang. 2024. CogAgent: A Visual Language Model for GUI Agents. arXiv:2312.08914 [cs.CV] https://arxiv.org/abs/2312.08914

    2024 arXiv

  5. [13]

    P ITU-T RECOMMENDATION. 1999. Subjective video quality assessment meth- ods for multimedia applications. (1999)

    1999

  6. [14]

    Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milanfar, and Feng Yang. 2021. MUSIQ: Multi-scale Image Quality Transformer. arXiv:2108.05997 [cs.CV] https://arxiv. org/abs/2108.05997

    2021 arXiv

  7. [15]

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimkühler, and George Drettakis

  8. [16]

    Arno Knapitsch, Jaesik Park, Qian-Yi Zhou, and Vladlen Koltun. 2017. Tanks and temples: Benchmarking large-scale scene reconstruction. ACM Transactions on Graphics (ToG) 36, 4 (2017), 1–13

    2017

  9. [17]

    Joo Chan Lee, Daniel Rho, Xiangyu Sun, Jong Hwan Ko, and Eunbyung Park

  10. [18]

    Feng Li, Renrui Zhang, Hao Zhang, Yuanhan Zhang, Bo Li, Wei Li, Zejun Ma, and Chunyuan Li. 2024. Llava-next-interleave: Tackling multi-image, video, and 3d in large multimodal models. arXiv preprint arXiv:2407.07895 (2024)

    2024 arXiv

  11. [19]

    Hanxue Liang, Tianhao Wu, Param Hanji, Francesco Banterle, Hongyun Gao, Rafal Mantiuk, and Cengiz Öztireli. 2024. Perceptual Quality Assessment of NeRF and Neural View Synthesis Methods for Front-Facing Views. In Computer Graphics Forum, Vol. 43. Wiley Online Library, e15036

    2024

  12. [20]

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024. Improved Baselines with Visual Instruction Tuning. arXiv:2310.03744 [cs.CV] https://arxiv.org/abs/ 2310.03744

    2024 arXiv

  13. [21]

    Xiangrui Liu, Xinju Wu, Pingping Zhang, Shiqi Wang, Zhu Li, and Sam Kwong

  14. [22]

    Haoyu Lu, Wen Liu, Bo Zhang, Bingxuan Wang, Kai Dong, Bo Liu, Jingxiang Sun, Tongzheng Ren, Zhuoshu Li, Hao Yang, Yaofeng Sun, Chengqi Deng, Hanwei Xu, Zhenda Xie, and Chong Ruan. 2024. DeepSeek-VL: Towards Real-World Vision-Language Understanding. arXiv:2403.05525 [cs.AI] htt...

    2024 arXiv

  15. [23]

    Tao Lu, Mulin Yu, Linning Xu, Yuanbo Xiangli, Limin Wang, Dahua Lin, and Bo Dai. 2024. Scaffold-GS: Structured 3D Gaussians for View-Adaptive Rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 20654–20664

    2024

  16. [24]

    In Proceedings of the 32nd ACM International Conference on Multimedia

    Compgs: Efficient 3d scene representation via compressed gaussian splat- ting. In Proceedings of the 32nd ACM International Conference on Multimedia . 2936–2944

  17. [25]

    Pedro Martin, António Rodrigues, João Ascenso, and Maria Paula Queluz. 2023. NeRF-QA: Neural Radiance Fields Quality Assessment Database. In 2023 15th International Conference on Quality of Multimedia Experience (QoMEX) . 107–110

    2023

  18. [26]

    Pedro Martin, António Rodrigues, João Ascenso, and Maria Paula Queluz. 2025. GS-QA: Comprehensive Quality Assessment Benchmark for Gaussian Splatting View Synthesis. arXiv preprint arXiv:2502.13196 (2025)

    2025 arXiv

  19. [27]

    Pedro Martin, António Rodrigues, João Ascenso, and Maria Paula Queluz. 2025. NeRF View Synthesis: Subjective Quality Assessment and Objective Metrics Evaluation. IEEE Access 13 (2025), 26–41

    2025

  20. [28]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. 2021. NeRF: representing scenes as neural radiance fields for view synthesis. Commun. ACM 65, 1 (2021), 99–106

    2021

  21. [29]

    Anish Mittal, Anush Krishna Moorthy, and Alan Conrad Bovik. 2012. No- reference image quality assessment in the spatial domain. IEEE Transactions on image processing 21, 12 (2012), 4695–4708

    2012

  22. [30]

    Meta. 2024. LLaMA 3.2: Revolutionizing Edge AI and Vision with Open, Cus- tomizable Models. https://ai.meta.com/blog/llama-3-2-revolutionizing-edge-ai- and-vision/. Meta AI Blog, retrieved December 2024

    2024

  23. [31]

    Simon Niedermayr, Josef Stumpfegger, and Rüdiger Westermann. 2024. Com- pressed 3D Gaussian Splatting for Accelerated Novel View Synthesis. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 10349–10358

    2024

  24. [32]

    Hamid R Sheikh and Alan C Bovik. 2006. Image information and visual quality. IEEE Transactions on image processing 15, 2 (2006), 430–444

    2006

  25. [33]

    KL Navaneet, Kossar Pourahmadi Meibodi, Soroush Abbasi Koohpayegani, and Hamed Pirsiavash. 2024. Compgs: Smaller and faster gaussian splatting with vector quantization. In European Conference on Computer Vision . Springer, 330– 349

    2024

  26. [34]

    Wei Sun, Xiongkuo Min, Danyang Tu, Siwei Ma, and Guangtao Zhai. 2023. Blind Quality Assessment for in-the-Wild Images via Hierarchical Feature Fusion and Iterative Mixed Database Training. IEEE Journal of Selected Topics in Signal Pro- cessing 17, 6 (Nov. 2023), 1178–1192. htt...

    2023

  27. [35]

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. 2004. Image quality assessment: from error visibility to structural similarity.IEEE transactions on image processing 13, 4 (2004), 600–612

    2004

  28. [36]

    Shaolin Su, Qingsen Yan, Yu Zhu, Cheng Zhang, Xin Ge, Jinqiu Sun, and Yanning Zhang. 2020. Blindly assess image quality in the wild guided by a self-adaptive hyper network. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 3667–3676

    2020

  29. [37]

    Zhou Wang, Eero P Simoncelli, and Alan C Bovik. 2003. Multiscale structural sim- ilarity for image quality assessment. In The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003 , Vol. 2. Ieee, 1398–1402

    2003

  30. [38]

    Haoning Wu, Zicheng Zhang, Weixia Zhang, Chaofeng Chen, Liang Liao, Chunyi Li, Yixuan Gao, Annan Wang, Erli Zhang, Wenxiu Sun, Qiong Yan, Xiongkuo Min, Guangtao Zhai, and Weisi Lin. 2023. Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels. arXiv:2312.17...

    2023 arXiv

  31. [39]

    Zhou Wang and Qiang Li. 2010. Information content weighting for perceptual image quality assessment. IEEE Transactions on image processing 20, 5 (2010), 1185–1198

    2010

  32. [40]

    Tianyi Xiong, Bo Li, Dong Guo, Huizhuo Yuan, Quanquan Gu, and Chun- yuan Li. 2024. LLaVA-OneVision-Chat: Improving Chat with Preference Learn- ing. https://github.com/LLaVA-VL/LLaVA-NeXT/blob/main/docs/LLaVA_ OneVision_Chat.md

    2024

  33. [41]

    Qi Yang, Kaifa Yang, Yuke Xing, Yiling Xu, and Zhu Li. 2024. A benchmark for gaussian splatting compression and quality assessment study. In Proceedings of the 6th ACM International Conference on Multimedia in Asia . 1–8. MM ’25, October 27–31, 2025, Dublin, Ireland Yuke Xing, et al

    2024

  34. [42]

    Yuke Xing, Qi Yang, Kaifa Yang, Yiling Xu, and Zhu Li. 2024. Explicit-NeRF-QA: A Quality Assessment Database for Explicit NeRF Model Compression. In 2024 IEEE International Conference on Visual Communications and Image Processing (VCIP). 1–5

    2024

  35. [43]

    Jiabo Ye, Haiyang Xu, Haowei Liu, Anwen Hu, Ming Yan, Qi Qian, Ji Zhang, Fei Huang, and Jingren Zhou. 2024. mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models. arXiv:2408.04840 [cs.CV] https://arxiv.org/abs/2408.04840

    2024 arXiv

  36. [44]

    Lin Zhang, Lei Zhang, Xuanqin Mou, and David Zhang. 2011. FSIM: A feature sim- ilarity index for image quality assessment. IEEE transactions on Image Processing 20, 8 (2011), 2378–2386

    2011

  37. [45]

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yujiu Yang. 2022. MANIQA: Multi-dimension Attention Net- work for No-Reference Image Quality Assessment. arXiv:2204.08958 [cs.CV] https://arxiv.org/abs/2204.08958

    2022 arXiv

  38. [46]

    Weixia Zhang, Kede Ma, Jia Yan, Dexiang Deng, and Zhou Wang. 2020. Blind Image Quality Assessment Using a Deep Bilinear Convolutional Neural Network. IEEE Transactions on Circuits and Systems for Video Technology 30, 1 (Jan. 2020), 36–47. https://doi.org/10.1109/tcsvt.2018.2886771

    2020

  39. [47]

    Weixia Zhang, Guangtao Zhai, Ying Wei, Xiaokang Yang, and Kede Ma. 2023. Blind image quality assessment via vision-language correspondence: A multitask learning perspective. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 14071–14081

    2023

  40. [48]

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang

  41. [52]

    Yuhang Zhang, Joshua Maraval, Zhengyu Zhang, Nicolas Ramin, Shishun Tian, and Lu Zhang. 2025. Evaluating Human Perception of Novel View Synthesis: Subjective Quality Assessment of Gaussian Splatting and NeRF in Dynamic Scenes. arXiv preprint arXiv:2501.08072 (2025)

    2025 arXiv

  42. [2018]

    In Proceedings of the IEEE conference on computer vision and pattern recognition

    The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition . 586–595

  43. [2023]

    ACM Transactions on Graphics 42, 4 (2023)

    3D Gaussian Splatting for Real-Time Radiance Field Rendering. ACM Transactions on Graphics 42, 4 (2023)

    2023

  44. [2024]

    arXiv preprint arXiv:2408.03822 (2024)

    Compact 3D Gaussian Splatting for Static and Dynamic Radiance Fields. arXiv preprint arXiv:2408.03822 (2024)

    2024 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.