Pith. sign in

REVIEW 2 major objections 5 minor 49 references

Semantic block matching plus selective high-frequency re-inpainting removes objects from 3D Gaussian scenes more coherently than one-shot methods.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 21:22 UTC pith:ASUFRTGR

load-bearing objection Solid engineering paper: two concrete modules (DINOv2 block matching + frequency-selective re-diffusion) give measurable gains over recent Gaussian inpainters; the match-correctness claim is under-validated but not fatal. the 2 major comments →

arxiv 2607.04144 v1 pith:ASUFRTGR submitted 2026-07-05 cs.RO

Semantic-Guided Progressive Object Removal with Gaussian Splatting

Xianliang Huang , Chen Xiao , Yuanxiang Ni , Guanming Liu , Mingkai Liu , Dikai Fan , Xiao Liu , Hao Zhang This is my paper
classification cs.RO
keywords 3D Gaussian Splattingobject removalsemantic block matchingDINOv2region-wise progressive refinementScore Distillation Samplingmulti-view inpaintinghigh-frequency guidance
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper claims that 3D object removal inside Gaussian Splatting becomes both sharper and more multi-view consistent when two steps replace the usual one-shot fill of an entire mask. First, DINOv2 embeddings of occluded blocks are matched by cosine similarity to the best-matching visible blocks from other camera views; those tokens are then injected via cross-attention into a diffusion model so the missing content is completed with semantically aligned material. Second, a simple high-frequency (Sobel) score flags only the sub-regions that still look blurry or inconsistent, and those blocks alone are re-diffused with an extra frequency prior. The authors show that the combination raises PSNR, SSIM and LPIPS inside the masked regions on forward-facing and unbounded scenes, while keeping training and memory costs modest relative to competing Gaussian and NeRF baselines. A reader who cares about AR/VR editing or robotics scene clean-up would therefore have a practical route to remove large objects without leaving ghosting or texture seams across views.

Core claim

The authors establish that Semantic-guided Block Matching (DINOv2 token retrieval followed by cross-attention injection) together with Region-Wise Progressive Refinement (high-frequency thresholding of sub-masks and selective re-diffusion) produces higher-fidelity, multi-view-coherent object removal inside 3D Gaussian Splatting than existing one-shot Gaussian-based or NeRF-based pipelines.

What carries the argument

Semantic-guided Block Matching (SBM) plus Region-Wise Progressive Refinement (RPR): occluded s imes s blocks are retrieved by DINOv2 cosine similarity and used as semantic tokens; low-frequency residual blocks are then selectively re-inpainted under a high-frequency prior.

Load-bearing premise

The method assumes that cosine similarity of DINOv2 block features reliably finds the geometrically correct source patch and that a simple Sobel magnitude threshold correctly marks only the blocks that still need re-inpainting.

What would settle it

On a held-out multi-view scene with large occlusions, disable either the DINOv2 matching or the high-frequency selective re-diffusion and measure whether the masked-region PSNR/SSIM/LPIPS gap versus the full pipeline disappears or reverses relative to the one-shot baselines reported in the paper.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper proposes a 3D object-removal pipeline built on 3D Gaussian Splatting that first obtains an initial completion via SDS from a latent diffusion model, then applies Semantic-guided Block Matching (SBM) and Region-Wise Progressive Refinement (RPR). SBM partitions the masked region into s imes s blocks, extracts DINOv2 features, retrieves the highest-cosine-similarity block from other views (Eqs. 8–10), and injects the matched token via cross-attention into the UNet. RPR flags low-quality blocks by a Sobel high-frequency magnitude threshold (15 % of surroundings) and re-diffuses only those blocks with both semantic and frequency conditioning (Eqs. 11–14). The refined multi-view images supervise the final Gaussian optimization under a combined L_GS + λ_SDS L_SDS loss. Quantitative results on SPIn-NeRF, self-captured, and Mip-NeRF 360 scenes (Table I) and ablations (Table III) show gains in masked-region PSNR/SSIM/LPIPS over SPIn-NeRF, GaussianEdit, GaussianGroup, MVInpainter and InFusion, with competitive runtime/memory (Table II).

Significance. If the reported gains hold under broader validation, the work supplies a practical, relatively efficient route to multi-view-consistent object removal inside the now-standard 3DGS representation—directly useful for AR/VR editing, robotics scene cleaning, and digital content pipelines. The explicit combination of DINOv2 block retrieval with selective high-frequency re-diffusion is a concrete engineering contribution beyond pure one-shot SDS or 2D inpainting baselines. Strengths include clear module ablations (Table III), multi-dataset quantitative tables restricted to masked pixels, and a favorable efficiency profile (7.9 h / 10 GB). The absence of match-accuracy diagnostics and statistical error bars limits the strength of the causal claim for SBM, but does not erase the empirical utility of the overall system.

major comments (2)
  1. [Section IV-B, Eqs. 8–10; Table III] Section IV-B (Eqs. 8–10) and the central claim of multi-view geometric coherence rest on the assumption that cosine similarity of DINOv2 features of s imes s blocks (τ = 0.1) selects geometrically and texturally correct source content. The manuscript never reports match accuracy, qualitative match visualizations, or failure cases under viewpoint change or repetitive texture. Ablation Table III only removes the entire SBM module; it does not isolate retrieval correctness from the mere addition of extra conditioning tokens. Without such evidence it remains possible that the PSNR/SSIM/LPIPS gains in Tables I–III arise from richer diffusion conditioning rather than correct cross-view alignment. A modest diagnostic (e.g., fraction of matches whose 3D re-projection lands inside the corresponding unmasked region, or a controlled ablation that replaces the best-match token by a random same-view
  2. [Sections IV-B.1, IV-C.1, V-A.1] The free parameters that control both modules—occlusion threshold τ = 0.1, high-frequency quality threshold δ = 15 % of surroundings, block size s = 15, and SDS weight/iteration count—are stated as empirical choices (Sections IV-B.1, IV-C.1, V-A.1) with no sensitivity analysis. Because the reported superiority over one-shot baselines depends on these thresholds, a short sensitivity study (or at least results for two neighboring values of τ and δ) is required to show that the gains are not brittle to the particular operating point chosen on the evaluation scenes.
minor comments (5)
  1. [Section IV-C.2, Fig. 1] Figure 1 caption and the surrounding text refer to a “High-frequency feature extractor” and “frequency-aware UNet,” yet the precise architecture of E_freq (Eq. 12) and how f_k is fused into the UNet (cross-attention vs. MLP) are never specified beyond a one-sentence description.
  2. [Table I] Table I reports point estimates without standard deviations or statistical significance tests across scenes; adding per-scene variance or a paired test would strengthen the quantitative claims.
  3. [Section IV-B] Notation for the semantic encoder switches between F_DINOv2 (Eq. 8) and F_sem (text after Eq. 8); a single consistent symbol would improve readability.
  4. [Section II-B] The abstract and introduction claim “outperforms existing Gaussian-based methods,” yet the related-work discussion of GScream and GaussianEditor is brief; a short paragraph clarifying the precise technical differences would help readers place the contribution.
  5. [Throughout] Minor typographical issues: “Region-Wise Progressive Refine- ment” line break, “Fe-map&mask” in Fig. 1, and inconsistent capitalization of “High-frequency.”

Circularity Check

0 steps flagged

No circularity: empirical engineering method whose claims rest on external baselines, ablations, and held-out metrics rather than any self-referential derivation.

full rationale

The paper proposes two modules (SBM via DINOv2 cosine matching + cross-attention, and RPR via Sobel high-frequency thresholding + selective re-diffusion) inside a 3DGS + SDS pipeline and evaluates them with standard image metrics (PSNR/SSIM/LPIPS) on public datasets (SPIn-NeRF, LLFF, Mip-NeRF 360) plus self-captured scenes against independent baselines (SPIn-NeRF, GaussianEditor, GaussianGroup, MVInpainter, InFusion). No equation equates a reported metric to a fitted free parameter by construction; the thresholds (τ=0.1, 15 % high-frequency) are acknowledged as empirical and are not used to “predict” the same quantities they were tuned on. Ablation Table III isolates each module by removal, confirming additive gains rather than tautological re-labeling. Self-citations to prior author work on NeRF inpainting appear only in Related Work and do not supply a uniqueness theorem, ansatz, or load-bearing premise for the new SBM/RPR claims. The derivation chain is therefore ordinary engineering: design → implement → measure against external references. Score 0 is the correct, non-manufactured outcome.

Axiom & Free-Parameter Ledger

4 free parameters · 3 axioms · 2 invented entities

The central claim rests on a handful of hand-chosen thresholds, the assumption that DINOv2 cosine similarity yields geometrically correct matches, and the standard 3DGS + SDS machinery. No new physical entities are postulated; the free parameters are purely algorithmic knobs.

free parameters (4)
  • occlusion ratio threshold τ = 0.1
    Blocks with mask overlap > τ are selected for matching; set empirically to 0.1 (Section IV-B). Directly controls which regions receive semantic guidance.
  • high-frequency quality threshold δ = 15 % of surroundings
    A block is refined if its Sobel magnitude is below 15 % of the surrounding area (Section IV-C). Determines which subregions enter the progressive stage.
  • block size s = 15
    Non-overlapping square blocks of side s; set to 15 (Implementation Details). Affects both matching granularity and compute.
  • SDS weight λ_SDS and iteration count = 1000 iterations (λ_SDS not numerically specified)
    Balances reconstruction loss against diffusion prior; 1000 SDS iterations used. Standard but still free hyperparameters that influence the initial inpainting quality.
axioms (3)
  • domain assumption DINOv2 features of masked blocks are sufficiently discriminative that cosine similarity recovers the correct multi-view correspondence for inpainting.
    Invoked throughout Section IV-B; no independent verification that the nearest neighbor is geometrically accurate rather than merely semantically similar.
  • domain assumption Score Distillation Sampling from a frozen Stable Diffusion v2.1 model supplies a useful 3D prior for Gaussian optimization after object removal.
    Standard in the cited SDS literature (DreamFusion et al.); used in Eq. (7) and the initial removal stage.
  • ad hoc to paper High-frequency Sobel energy is a reliable proxy for perceptual quality of an inpainted block.
    Introduced in Section IV-C to decide which blocks enter RPR; not derived from a perceptual study.
invented entities (2)
  • Semantic-guided Block Matching (SBM) no independent evidence
    purpose: Retrieve and inject multi-view semantic tokens to condition diffusion inpainting of occluded blocks.
    Core algorithmic contribution of the paper; no independent existence outside this work.
  • Region-Wise Progressive Refinement (RPR) no independent evidence
    purpose: Selectively re-diffuse only low high-frequency blocks to improve local fidelity without full re-generation.
    Second core module; defined and evaluated only inside this paper.

pith-pipeline@v1.1.0-grok45 · 17601 in / 2954 out tokens · 26289 ms · 2026-07-11T21:22:15.060341+00:00 · methodology

0 comments
read the original abstract

Removing unwanted objects from reconstructed 3D scenes is an important task in computer vision, supporting applications in AR/VR, robotics, and digital content creation. Existing methods typically complete the entire masked region in a single step and without effectively utilizing semantic information from other views, leading to difficulties in handling complex geometric details and textures. In this work, we propose a novel framework that integrates Semantic-guided Block Matching (SBM) and Region-Wise Progressive Refinement (RPR) for high-quality 3D object removal. First, we leverage DINOv2 to encode semantic guidance from multi-view observations, and the best match tokens are decoded to complete missing regions in the target view while maintaining cross-view consistency. Second, we introduce a RPR strategy that segments the target mask into multiple subregions and selectively refines those with poor visual quality. Our method is built upon Gaussian Splatting, ensuring high-fidelity scene reconstruction with efficient computation. Experimental results demonstrate that our approach outperforms existing Gaussian-based methods in terms of perceptual quality and coherence in 3D object removal.

Figures

Figures reproduced from arXiv: 2607.04144 by Chen Xiao, Dikai Fan, Guanming Liu, Hao Zhang, Mingkai Liu, Xianliang Huang, Xiao Liu, Yuanxiang Ni.

Figure 1
Figure 1. Figure 1: Overview of our pipeline. The key insight of our method is to iteratively refine all region blocks in a few steps, guided by the high-frequency information from the semantic-guided block. Specifically, the semantic-guided block matching leverages multi-view observations to guide the completion of missing regions with content-aligned semantics. Next, the high-frequency feature extractor provides auxiliary s… view at source ↗
Figure 2
Figure 2. Figure 2: Given a 3D scene, our method produces accurate object removal [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative results. We compare our method with SPIn-NeRF and GaussianEdit in a consistent view. TABLE I QUANTITATIVE COMPARISONS WITH GAUSSIAN-BASED METHODS ON SPIN-NERF, SELF-CAPTURED, AND MIP-NERF 360 DATASETS. Method SPIn-NeRF dataset Self-captured dataset Mip-NeRF 360 dataset PSNR ↑ SSIM ↑ LPIPS ↓ PSNR ↑ SSIM ↑ LPIPS ↓ PSNR ↑ SSIM ↑ LPIPS ↓ SPIn-NeRF 26.8 0.901 0.176 21.7 0.824 0.231 23.0 0.860 0.204 … view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results on Bear and Bicycle scenes. Given the input images, we visualize the removal results of MVInpainter, GaussianGroup, InFusion and our method in a consistent view. TABLE III ABLATION STUDY. WE REPORT THE AVERAGE METRICS ON THE SELF-CAPTURED SCENE AND THE PUBLIC SCENE. Method Self-captured scene Public scene PSNR↑ SSIM↑ LPIPS↓ PSNR↑ SSIM↑ LPIPS↓ w/o SBM 24.58 0.681 0.312 24.18 0.605 0.369 … view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 11 linked inside Pith

  1. [1]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoor- thi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,”Communications of the ACM, vol. 65, no. 1, pp. 99–106, 2021

    2021

  2. [2]

    3d gaussian splatting for real-time radiance field rendering

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering.”ACM Trans. Graph., vol. 42, no. 4, pp. 139–1, 2023

    2023

  3. [3]

    Live-gs: Llm powers interactive vr by enhancing gaussian splatting,

    H. Mao, Z. Xu, S. Wei, Y . Quan, N. Deng, and X. Yang, “Live-gs: Llm powers interactive vr by enhancing gaussian splatting,”arXiv preprint arXiv:2412.09176, 2024

    Pith/arXiv arXiv 2024

  4. [4]

    Mace: Mixture-of-experts accelerated coordinate encoding for large-scale scene localization and rendering,

    M. Liu, D. Fan, H. Que, H. Gao, X. Liu, S. Peng, M. Lin, S. Gu, R. Ye, W. Qiu,et al., “Mace: Mixture-of-experts accelerated coordinate encoding for large-scale scene localization and rendering,”arXiv preprint arXiv:2510.14251, 2025

    arXiv 2025

  5. [5]

    Mind-v: Hierarchical video generation for long-horizon robotic manipulation with rl-based physical alignment,

    R. Zhang, M. Zhang, J. Zhou, Z. Guo, X. Liu, Z. Xu, Z. Zhong, P. Yan, H. Luo, and X. Li, “Mind-v: Hierarchical video generation for long-horizon robotic manipulation with rl-based physical alignment,” arXiv preprint arXiv:2512.06628, 2025

    Pith/arXiv arXiv 2025

  6. [6]

    Drivinggaussian: Composite gaussian splatting for surrounding dy- namic autonomous driving scenes,

    X. Zhou, Z. Lin, X. Shan, Y . Wang, D. Sun, and M.-H. Yang, “Drivinggaussian: Composite gaussian splatting for surrounding dy- namic autonomous driving scenes,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024, pp. 21 634–21 643

    2024

  7. [7]

    Spin-nerf: Multiview segmentation and perceptual inpainting with neural radiance fields,

    A. Mirzaei, T. Aumentado-Armstrong, K. G. Derpanis, J. Kelly, M. A. Brubaker, I. Gilitschenski, and A. Levinshtein, “Spin-nerf: Multiview segmentation and perceptual inpainting with neural radiance fields,” arXiv preprint arXiv:2211.12254, 2022

    Pith/arXiv arXiv 2022

  8. [8]

    Gaussianeditor: Swift and controllable 3d editing with gaussian splatting,

    Y . Chen, Z. Chen, C. Zhang, F. Wang, X. Yang, Y . Wang, Z. Cai, L. Yang, H. Liu, and G. Lin, “Gaussianeditor: Swift and controllable 3d editing with gaussian splatting,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 476–21 485

    2024

  9. [9]

    Nerf-mir: Toward high-quality restoration of masked images with neural radiance fields,

    X. Huang, Z. Zhong, S. Chen, Y . Xu, J. Guan, and S. Zhou, “Nerf-mir: Toward high-quality restoration of masked images with neural radiance fields,”IEEE Transactions on Neural Networks and Learning Systems, 2026

    2026

  10. [10]

    Mvip-nerf: Multi-view 3d inpainting on nerf scenes via diffusion prior,

    H. Chen, C. C. Loy, and X. Pan, “Mvip-nerf: Multi-view 3d inpainting on nerf scenes via diffusion prior,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 5344–5353

    2024

  11. [11]

    Infusion: Inpainting 3d gaussians via learning depth completion from diffusion prior,

    Z. Liu, H. Ouyang, Q. Wang, K. L. Cheng, J. Xiao, K. Zhu, N. Xue, Y . Liu, Y . Shen, and Y . Cao, “Infusion: Inpainting 3d gaussians via learning depth completion from diffusion prior,”arXiv preprint arXiv:2404.11613, 2024

    Pith/arXiv arXiv 2024

  12. [12]

    Generative object insertion in gaussian splatting with a multi-view diffusion model,

    H. Zhong, C. Wang, J. Zhang, and J. Liao, “Generative object insertion in gaussian splatting with a multi-view diffusion model,”Visual Informatics, p. 100238, 2025

    2025

  13. [13]

    Dreamfusion: Text- to-3d using 2d diffusion,

    B. Poole, A. Jain, J. T. Barron, and B. Mildenhall, “Dreamfusion: Text- to-3d using 2d diffusion,”arXiv preprint arXiv:2209.14988, 2022

    Pith/arXiv arXiv 2022

  14. [14]

    Seeing the unseen: A frequency prompt guided transformer for image restoration,

    S. Zhou, J. Pan, J. Shi, D. Chen, L. Qu, and J. Yang, “Seeing the unseen: A frequency prompt guided transformer for image restoration,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 246–264

    2024

  15. [15]

    Devil is in the uniformity: Exploring diverse learners within transformer for image restoration,

    S. Zhou, D. Li, J. Pan, J. Zhou, J. Shi, and J. Yang, “Devil is in the uniformity: Exploring diverse learners within transformer for image restoration,” inProc. Int. Conf. Comput. Vis., 2025

    2025

  16. [16]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,”Advances in neural information processing systems, vol. 33, pp. 6840–6851, 2020

    2020

  17. [17]

    Iddr-ngp: Incorporating detectors for distractors removal with instant neural radiance field,

    X. Huang, J. Gou, S. Chen, Z. Zhong, J. Guan, and S. Zhou, “Iddr-ngp: Incorporating detectors for distractors removal with instant neural radiance field,” inProceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 1343–1351

    2023

  18. [18]

    High-resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High-resolution image synthesis with latent diffusion models,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 684–10 695

    2022

  19. [19]

    Learning an adaptive sparse transformer for efficient image restoration,

    S. Zhou, J. Pan, and J. Yang, “Learning an adaptive sparse transformer for efficient image restoration,”IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 47, no. 11, pp. 10 344–10 360, 2025

    2025

  20. [20]

    Text-driven image editing via learnable regions,

    Y . Lin, Y .-W. Chen, Y .-H. Tsai, L. Jiang, and M.-H. Yang, “Text-driven image editing via learnable regions,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 7059–7068

    2024

  21. [21]

    Adding conditional control to text-to-image diffusion models,

    L. Zhang, A. Rao, and M. Agrawala, “Adding conditional control to text-to-image diffusion models,” 2023

    2023

  22. [22]

    Spatext: Spatio-textual repre- sentation for controllable image generation,

    O. Avrahami, T. Hayes, O. Gafni, S. Gupta, Y . Taigman, D. Parikh, D. Lischinski, O. Fried, and X. Yin, “Spatext: Spatio-textual repre- sentation for controllable image generation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 18 370–18 380

    2023

  23. [23]

    Dreambooth: Fine tuning text-to-image diffusion models for subject- driven generation,

    N. Ruiz, Y . Li, V . Jampani, Y . Pritch, M. Rubinstein, and K. Aberman, “Dreambooth: Fine tuning text-to-image diffusion models for subject- driven generation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 22 500–22 510

    2023

  24. [24]

    An image is worth one word: Personal- izing text-to-image generation using textual inversion,

    R. Gal, Y . Alaluf, Y . Atzmon, O. Patashnik, A. H. Bermano, G. Chechik, and D. Cohen-Or, “An image is worth one word: Personal- izing text-to-image generation using textual inversion,”arXiv preprint arXiv:2208.01618, 2022

    Pith/arXiv arXiv 2022

  25. [25]

    Sdedit: Guided image synthesis and editing with stochastic differen- tial equations,

    C. Meng, Y . He, Y . Song, J. Song, J. Wu, J.-Y . Zhu, and S. Ermon, “Sdedit: Guided image synthesis and editing with stochastic differen- tial equations,”arXiv preprint arXiv:2108.01073, 2021

    Pith/arXiv arXiv 2021

  26. [26]

    Diffusion models beat gans on image synthesis,

    P. Dhariwal and A. Nichol, “Diffusion models beat gans on image synthesis,”Advances in neural information processing systems, vol. 34, pp. 8780–8794, 2021

    2021

  27. [27]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,”arXiv preprint arXiv:2010.02502, 2020

    Pith/arXiv arXiv 2010

  28. [28]

    Deepsdf: Learning continuous signed distance functions for shape representation,

    J. J. Park, P. Florence, J. Straub, R. Newcombe, and S. Lovegrove, “Deepsdf: Learning continuous signed distance functions for shape representation,” inProceedings of the IEEE/CVF conference on com- puter vision and pattern recognition, 2019, pp. 165–174

    2019

  29. [29]

    Shape inpainting using 3d generative adversarial network and recurrent convolutional networks,

    W. Wang, Q. Huang, S. You, C. Yang, and U. Neumann, “Shape inpainting using 3d generative adversarial network and recurrent convolutional networks,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 2298–2306

    2017

  30. [30]

    Textde- former: Geometry manipulation using text guidance,

    W. Gao, N. Aigerman, T. Groueix, V . Kim, and R. Hanocka, “Textde- former: Geometry manipulation using text guidance,” inACM SIG- GRAPH 2023 conference proceedings, 2023, pp. 1–11

    2023

  31. [31]

    Clip-nerf: Text-and- image driven manipulation of neural radiance fields,

    C. Wang, M. Chai, M. He, D. Chen, and J. Liao, “Clip-nerf: Text-and- image driven manipulation of neural radiance fields,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition, 2022, pp. 3835–3844

    2022

  32. [32]

    Dreameditor: Text- driven 3d scene editing with neural fields,

    J. Zhuang, C. Wang, L. Lin, L. Liu, and G. Li, “Dreameditor: Text- driven 3d scene editing with neural fields,” inSIGGRAPH Asia 2023 Conference Papers, 2023, pp. 1–10

    2023

  33. [33]

    Hi-nerf: Hybridizing 2d inpainting with neural radiance fields for 3d scene inpainting,

    X. Huang, S. Chen, Z. Zhong, J. Gou, J. Guan, and S. Zhou, “Hi-nerf: Hybridizing 2d inpainting with neural radiance fields for 3d scene inpainting,” inProceedings of the Asian Conference on Computer Vision, 2024, pp. 2855–2871

    2024

  34. [34]

    Text2mesh: Text-driven neural stylization for meshes,

    O. Michel, R. Bar-On, R. Liu, S. Benaim, and R. Hanocka, “Text2mesh: Text-driven neural stylization for meshes,” inProceedings of the IEEE/CVF conference on computer vision and pattern recogni- tion, 2022, pp. 13 492–13 502

    2022

  35. [35]

    Nerfiller: Completing scenes via generative 3d inpainting,

    E. Weber, A. Holynski, V . Jampani, S. Saxena, N. Snavely, A. Kar, and A. Kanazawa, “Nerfiller: Completing scenes via generative 3d inpainting,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 20 731–20 741

    2024

  36. [36]

    Instant3dit: Multiview inpainting for fast editing of 3d objects,

    A. Barda, M. Gadelha, V . G. Kim, N. Aigerman, A. H. Bermano, and T. Groueix, “Instant3dit: Multiview inpainting for fast editing of 3d objects,”arXiv preprint arXiv:2412.00518, 2024

    Pith/arXiv arXiv 2024

  37. [37]

    Progressive3d: Progressively local editing for text-to-3d content creation with complex semantic prompts,

    X. Cheng, T. Yang, J. Wang, Y . Li, L. Zhang, J. Zhang, and L. Yuan, “Progressive3d: Progressively local editing for text-to-3d content creation with complex semantic prompts,”arXiv preprint arXiv:2310.11784, 2023

    Pith/arXiv arXiv 2023

  38. [38]

    Gaussian grouping: Segment and edit anything in 3d scenes,

    M. Ye, M. Danelljan, F. Yu, and L. Ke, “Gaussian grouping: Segment and edit anything in 3d scenes,” inEuropean Conference on Computer Vision. Springer, 2025, pp. 162–179

    2025

  39. [39]

    Learning 3d geometry and feature consistent gaussian splatting for object removal,

    Y . Wang, Q. Wu, G. Zhang, and D. Xu, “Learning 3d geometry and feature consistent gaussian splatting for object removal,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 1–17

    2024

  40. [40]

    Mvinpainter: Learning multi-view consistent inpainting to bridge 2d and 3d editing,

    C. Cao, C. Yu, F. Wang, X. Xue, and Y . Fu, “Mvinpainter: Learning multi-view consistent inpainting to bridge 2d and 3d editing,”arXiv preprint arXiv:2408.08000, 2024

    Pith/arXiv arXiv 2024

  41. [41]

    Ewa volume splatting,

    M. Zwicker, H. Pfister, J. Van Baar, and M. Gross, “Ewa volume splatting,” inProceedings Visualization, 2001. VIS’01.IEEE, 2001, pp. 29–538

    2001

  42. [42]

    Optical models for direct volume rendering,

    N. Max, “Optical models for direct volume rendering,”TVCG, vol. 1, no. 2, pp. 99–108, 1995

    1995

  43. [43]

    Resolution-robust large mask inpainting with fourier convolutions,

    R. Suvorov, E. Logacheva, A. Mashikhin, A. Remizova, A. Ashukha, A. Silvestrov, N. Kong, H. Goka, K. Park, and V . Lempitsky, “Resolution-robust large mask inpainting with fourier convolutions,” inProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2022, pp. 2149–2159

    2022

  44. [44]

    Structure-from-motion revisited,

    J. L. Schonberger and J.-M. Frahm, “Structure-from-motion revisited,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 4104–4113

    2016

  45. [45]

    Design of an image edge detection filter using the sobel operator,

    N. Kanopoulos, N. Vasanthavada, and R. L. Baker, “Design of an image edge detection filter using the sobel operator,”IEEE Journal of solid-state circuits, vol. 23, no. 2, pp. 358–367, 1988

    1988

  46. [46]

    Image quality assessment: from error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,”TIP, vol. 13, no. 4, pp. 600–612, 2004

    2004

  47. [47]

    Local light field fusion: Practical view synthesis with prescriptive sampling guidelines,

    B. Mildenhall, P. P. Srinivasan, R. Ortiz-Cayon, N. K. Kalantari, R. Ramamoorthi, R. Ng, and A. Kar, “Local light field fusion: Practical view synthesis with prescriptive sampling guidelines,”ACM Transactions on Graphics (TOG), 2019

    2019

  48. [48]

    Mip-nerf 360: Unbounded anti-aliased neural radiance fields,

    J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5470–5479

    2022

  49. [49]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586–595

    2018