Pith. sign in

REVIEW 4 major objections 5 minor 40 references

3D-Consistent Image Inpainting with Diffusion Models

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Training a diffusion model on paired views lets it inpaint occluded regions that stay consistent with the scene's 3D structure — without 3D supervision.

desk verdict The second-viewpoint conditioning idea is genuinely useful, but the evaluation never defines what the metrics are measured against, so the '3D-consistent outperformance' claim is not yet supported. read the letter →

arxiv 2412.05881 v1 pith:FBZJM5HH submitted 2024-12-08 cs.CV cs.AI

Leonid Antsfeld , Boris Chidlovskii This is my paper · ORCID
classification cs.CVcs.AI
keywords imageinpaintingdiffusionmodels3Dconsistencyin-contextguidancemulti-viewtrainingunconditionalDDPMcross-viewcompletionnoisescheduleresampling
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

The paper tries to establish that an unconditional diffusion model, trained only on pairs of images showing the same scene, learns enough 3D structure to repaint masked occluders in a geometrically consistent way. The claim is that injecting a clean second viewpoint of the scene into the denoising process acts as in-context guidance, harmonizing masked and unmasked regions while keeping the result coherent with the scene's actual geometry. If true, this gives a route to 3D-aware inpainting that needs no 3D labels, no depth maps, and no multi-view geometry at inference — just a second picture of the same place. The authors report that their method, InConDiff, beats RePaint, Stable Diffusion, and Stable Diffusion XL on four datasets with both semantic and random masks.

What carries the argument

The key machinery is in-context guidance via cross-view completion: a Diffusion Transformer whose decoder cross-attends to clean image patches from a second view of the same scene, trained with the standard denoising objective $L_2 = \mathbb{E}_{t,x_0,\epsilon}\|\epsilon - g_\theta(x_t,t,x')\|^2$. The second view provides the inductive bias that lets a 2D denoiser absorb 3D priors. This is paired with RePaint-style post-conditioning of the reverse process, where known pixels are resampled from the forward noising distribution while masked pixels come from the model, and a Laplace noise schedule that reduces the number of resampling jumps needed for boundary harmonization.

What would settle it

Open Section 4 and the authors' released code and check which image is used as the reference for PSNR, LPIPS, and SSIM. If the scores compare against the original masked image that still contains the occluder, the metrics penalize exactly the desired removal; on HM3D, one could render clean background images with known geometry and recompute the numbers, which would settle whether the advantage over Stable Diffusion XL reflects true 3D consistency.

Watch

Extended reading notes

Core claim

The central discovery is that 3D consistency is not something a diffusion model has to be given; it can be induced by training the denoiser on pairs of views. InConDiff modifies an unconditional DDPM so that the denoising network sees, in addition to the noisy image being denoised, a clean image from an alternative viewpoint of the same scene. The decoder cross-attends to the clean view's tokens while predicting the noise, letting it compare the two views and recover geometric cues during 2D denoising. At inference, a mask is applied post hoc, and a resampling schedule with noise jumps harmonizes the boundary between generated and known pixels. The authors argue that this yields semantically coherent and 3D-consistent inpaintings without explicit 3D supervision.

Load-bearing premise

The claim depends on the assumption that a second viewpoint of the same scene injects enough 3D structure into an otherwise 2D denoiser to make inpaintings geometrically consistent, and that the reported metrics compare against a ground truth the paper never actually defines.

Editorial extensions

If this is right

  • If the claim holds, inpainting systems can be made 3D-aware by training on any paired-view corpus, including photo collections with known camera poses, rather than requiring synthetic 3D data or depth supervision.
  • Post-hoc masking means the same trained model can handle novel mask shapes, including masks that cover a large fraction of the image, without retraining on masks.
  • The reported performance gap over Stable Diffusion XL suggests that scene-specific geometry, not just large-scale 2D priors, is the limiting factor for consistent content filling.
  • A faster Laplace-based resampling schedule cuts the inference-time overhead of harmonization, making the approach more practical for real use.

Reading between the lines

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

  • If the method scales to video, the same in-context principle suggests that using several frames rather than one additional image could remove moving objects consistently across time — the paper itself names multi-view extension as future work.
  • The metric protocol is an open question: the paper never states whether PSNR, LPIPS, and SSIM are computed against the original image with the occluder or against a clean reference, so the reported gains should be read with that ambiguity in mind.
  • A natural testable extension is single-image inpainting by retrieving a second view from a visual database, making the in-context image a retrieved prior rather than a required input.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes InConDiff, an image-inpainting method that extends an unconditional DiT-based diffusion model by injecting a second image of the same scene as in-context guidance during denoising. The authors argue that the cross-view information acts as a 3D prior, allowing harmonized and 3D-consistent inpainting without explicit 3D supervision. They describe a post-conditioning scheme with resampling and a Laplace noise schedule, and report quantitative results on HM3D, MegaDepth, StreetView, and WalkingTour, comparing against RePaint, Stable Diffusion, and Stable Diffusion XL. The central claim is that InConDiff outperforms the state of the art and "ensures the 3D consistency" of inpainted regions.

Significance. If the claims were fully supported, the method would be a simple and practical way to inject geometric awareness into 2D diffusion inpainting, with potential applications in object removal, view synthesis, and robotics. The architectural modification is clean and the paper provides code, ablations on mask ratio and resampling schedule, and a transparent comparison to several baselines. However, the evidence as presented is not yet load-bearing: the evaluation never defines the reference image for PSNR/LPIPS/SSIM, no metric operationalizes "3D consistency", and key numerical results appear internally inconsistent. These gaps directly affect the abstract's claim of 3D-consistent outperformance, so the paper requires substantial revision before the quantitative conclusions can be accepted.

major comments (4)
  1. [Section 4 (Evaluation Metrics); Tables 1-3] The reference image for PSNR, LPIPS, and SSIM is never specified. The task is to remove occluders (cars, pedestrians, poles), so the original unmasked image still contains the very content to be erased; computing metrics against that image would penalize correct removal. If the metrics are computed against a second viewpoint warped into the target view, a manually cleaned background, or a reconstructed occluder-free image, that protocol is absent. The paper must state the exact reference and the region over which metrics are averaged, and report results separately for the masked area if relevant. Without this, the reported advantage over SD/SD-XL is uninterpretable.
  2. [Abstract; Section 4; Conclusion] No metric operationalizes "3D consistency", which is the paper's central claim. Tables 1-4 report only 2D image-similarity metrics. The authors should add a quantitative geometric measure, for example: (i) multi-view reprojection error between the inpainted image and a second view, using known depth/pose; (ii) photometric or feature-metric consistency after warping the inpainted region to the second view; (iii) depth-map consistency with a pretrained monocular depth estimator before and after inpainting; or (iv) a downstream geometric task such as relative pose estimation accuracy on inpainted images. Without such a metric, the phrase "ensures the 3D consistency" is not verified by the reported experiments.
  3. [Table 4] The PSNR values for Njumps=1 are 18.7 regardless of Nsteps (250, 500, 1000), and LPIPS and SSIM are also identical across those columns (0.26 and 0.83 respectively). This suggests either a copy-paste or alignment error in the table, or that the number of denoising steps has no effect on output quality when no resampling is used, which would be surprising for a DDPM. The row should be rechecked and corrected. Additionally, no variance or confidence intervals are reported for any table, so the significance of small differences from baselines (e.g., MegaDepth semantic SSIM 0.86 vs 0.88) cannot be assessed.
  4. [Section 4.1 (Comparison to State-of-the-Art)] The comparison protocol is underspecified. The paper does not state the number of test images per dataset, how semantic and random masks are generated (beyond the random-mask description), whether the exact same masks are used for all compared methods, or whether the Stable Diffusion and SD-XL checkpoints are fine-tuned on the target datasets. These details are necessary to interpret the quantitative claims, especially because the baselines are general-purpose models and may not have been adapted to the dataset-specific distribution. The authors should provide the full evaluation setup, ideally in the main text or a clearly marked supplementary document.
minor comments (5)
  1. [Section 1 (Contributions)] The phrase "We prove its efficiency of InConDiff" is ungrammatical and should read "We demonstrate the effectiveness of InConDiff" or similar.
  2. [Section 3.3 (Conditioning DDPMs on the known region)] The notation for the noise schedule is inconsistent: Equation (8) uses \hat\alpha_t while Equations (1)-(2) use \bar\alpha_t. Please unify the notation.
  3. [Figure 6 caption] The caption says "semantic (top) and random (top) masks" where the second "top" should presumably be "bottom".
  4. [Section 2 (Related Work)] There is a typo in the reference to Sohl-Dickstein et al.; the name is written as "SohlDickstein" without a space.
  5. [General] The term "in-context" is used throughout but never formally defined. Since "in-context learning" has a well-known meaning in large language models, the authors should clarify that here it refers to conditioning on an additional image as input to the denoising network.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's derivation chain is self-contained; self-citations to CroCo/CroCo v2 are real, code-reproduced evidence and no target quantity is defined in terms of a fitted parameter or the paper's own outputs.

full rationale

The derivation chain is standard DDPM mathematics extended with a second image as input. Equations (1)-(6) give the ordinary forward/reverse diffusion process and the noise-prediction objective; Eq. (7) simply adds the in-context image x' as an extra argument to the denoiser; Eq. (8) is the RePaint-style post-conditioning that combines model samples for masked regions with forward-diffused known regions. None of these equations define the claimed 3D-consistent output in terms of the evaluation metrics, nor is any parameter fitted to the reported PSNR/LPIPS/SSIM values. The Laplace noise schedule and the resampling steps/jumps are chosen by validation and ablation (Tables 3-4), which is hyperparameter tuning rather than a fitted-input-called-prediction pattern. The only self-citations are [28,29] for the cross-view completion architecture, pair-quality scoring, and ViT dimensions (Section 3.2 and Implementation Details). These citations are not circularly load-bearing: CroCo/CroCo v2 are code-reproduced and externally benchmarked on stereo and optical flow, and the present paper trains its own model from scratch and compares against external baselines (RePaint, SD, SD-XL) on external datasets. The under-specified evaluation reference for PSNR/LPIPS/SSIM is a real rigor concern about what the metrics compare against, but it is a correctness/interpretability limitation, not a circularity: it does not make any reported number equal to an input by construction. The Limitations section even concedes that in crowded scenes the inpainting 'can replace masked obstacles with other obstacles,' which is an honest scope restriction rather than a self-validating claim. Overall, no circular step is identifiable in the paper's equations, claims, or citation chain.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The method's central premise is that a second image of the same scene, injected as in-context conditioning during diffusion training and inference, transfers 3D information without explicit 3D supervision. The paper does not prove this; it treats it as an inductive-bias assumption borrowed from cross-view completion. No new physical entities are introduced.

free parameters (3)
  • Laplace noise schedule = undefined; adopted after experiments
    Section 3.3 states the Laplace schedule is adopted based on experiments, but no equation or parameters are given.
  • Resampling steps and jumps = 1000 steps, 10 jumps
    Section 4 (Implementation details) specifies these values; ablation Table 4 selects them.
  • Image pair selection thresholds = unspecified
    Section 4 says pairs are scored by overlap and viewpoint angle following CroCo v2, but exact thresholds are not reported.
assumptions (5)
  • standard math The DDPM forward and reverse processes, and the RePaint resampling update, are correct as stated.
    Section 3.1 relies on the standard DDPM formulation from Ho et al. (2020).
  • domain assumption Cross-view completion training transfers geometric cues that are useful for 3D tasks.
    Section 3.2 adopts the CroCo principle that cross-view completion enables perception of low-level geometric cues.
  • domain assumption The second image x' provides informative content for the masked areas at test time.
    Section 4 describes pair selection by overlap and viewpoint angle, but the paper does not analyze failure cases where the second view misses the occluded content.
  • domain assumption DINOv2-based occlusion detection correctly identifies the regions to mask.
    Section 3.3 uses DINOv2 to detect vehicle and pedestrian classes; no accuracy analysis is provided.
  • ad hoc to paper The Laplace noise schedule improves boundary harmonization and halves overhead.
    Section 3.3 states this without defining the schedule or showing supporting experiments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 3D-Consistent Image Inpainting with Diffusion Models." pith.science (2026). https://pith.science/paper/FBZJM5HH

@misc{pith2026241205881,
  author       = {Pith},
  title        = {Pith review of: 3D-Consistent Image Inpainting with Diffusion Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FBZJM5HH}},
  note         = {Machine review of arXiv:2412.05881}
}
read the original abstract

We address the problem of 3D inconsistency of image inpainting based on diffusion models. We propose a generative model using image pairs that belong to the same scene. To achieve the 3D-consistent and semantically coherent inpainting, we modify the generative diffusion model by incorporating an alternative point of view of the scene into the denoising process. This creates an inductive bias that allows to recover 3D priors while training to denoise in 2D, without explicit 3D supervision. Training unconditional diffusion models with additional images as in-context guidance allows to harmonize the masked and non-masked regions while repainting and ensures the 3D consistency. We evaluate our method on one synthetic and three real-world datasets and show that it generates semantically coherent and 3D-consistent inpaintings and outperforms the state-of-art methods.

Figures

Figures reproduced from arXiv: 2412.05881 by the authors.

Figure 1
Figure 1. 3D-consistent inpainting with InConDiff: a) original image, b) image with masked occlusions; c) in-context image; d) InConDiff inpainting preserving 3D consistency. [1,10,36]. Their inability to control the 3D properties of objects in the generated images is particularly harm￾ful for image inpainting, which aims to harmonizing 3D properties of masked and non-masked regions. 3D understanding of diffusion models has b… view at source ↗
Figure 3
Figure 3. ViT architecture for learning the denoising [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Image pairs for training unconditional DDPMs from HM3D, MegaDepth, StreetView and WalkingTour [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: Inpainting with semantic mask using noise schedule with jumps. Note resampling effect for harmoniza [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: StreetView dataset: inpainting with semantic [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 8
Figure 8. Figure 8: HM3D dataset: inpainting with semantic (top) and random (bottom) masks. Origin Mask In-context Inpainting [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: WalkingTour: inpainting with semantic (top) [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 24 canonical work pages

  1. [1]

    Mitra, and Paul Guer- rero

    Titas Anciukevicius, Zexiang Xu, Matthew Fisher, Paul Henderson, Hakan Bilen, Niloy J. Mitra, and Paul Guer- rero. Renderdiffusion: Image diffusion for 3d recon- struction, inpainting and generation. InIEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 12608–12618. IEEE, 2023. 1 8 3D-Consistent Image Inpainting with Diffusion Models

    work page 2023

  2. [2]

    Patchmatch: A randomized corre- spondence algorithm for structural image editing.ACM Trans

    Connelly Barnes, Eli Shechtman, Adam Finkelstein, and Dan B Goldman. Patchmatch: A randomized corre- spondence algorithm for structural image editing.ACM Trans. Graph., 28(3):24, 2009. 1

    work page 2009

  3. [3]

    Conditional image gen- eration with score-based diffusion models

    Georgios Batzolis, Jan Stanczuk, Carola-Bibiane Schön- lieb, and Christian Etmann. Conditional image gen- eration with score-based diffusion models. CoRR, abs/2111.13606, 2021. 2

    arXiv 2021

  4. [4]

    On the importance of noise scheduling for diffusion models.CoRR, abs/2301.10972, 2023

    Ting Chen. On the importance of noise scheduling for diffusion models.CoRR, abs/2301.10972, 2023. 4

    arXiv 2023

  5. [5]

    Latentpaint: Image inpainting in latent space with diffusion models

    CiprianCorneanu,RaghudeepGadde,andAleixM.Mar- tinez. Latentpaint: Image inpainting in latent space with diffusion models. InIEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 4334– 4343, 2024. 1, 2, 6

    work page 2024

  6. [6]

    Diffusion mod- els beat gans on image synthesis

    Prafulla Dhariwal and Alex Nichol. Diffusion mod- els beat gans on image synthesis. arXiv preprint arXiv:2105.05233, 2021. 1

    arXiv 2021

  7. [7]

    Improved noise sched- ule for diffusion training.arXiv:2407.03297, 2024

    Tiankai Hang and Shuyang Gu. Improved noise sched- ule for diffusion training.arXiv:2407.03297, 2024. 4

    arXiv 2024

  8. [8]

    Classifier-free diffusion guidance, 2022

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance, 2022. 2

    2022

Show all 40 references
  1. [9]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. InNeurIPS, 2020. 1, 2, 3

    2020

  2. [10]

    Diffusionmodel-basedimage editing: A survey.arXiv:2402.17525, 2024

    Yi Huang, Jiancheng Huang, Yifan Liu, Mingfu Yan, Jiaxi Lv, Jianzhuang Liu, Wei Xiong, He Zhang, Shifeng Chen,andLiangliangCao. Diffusionmodel-basedimage editing: A survey.arXiv:2402.17525, 2024. 1

    2024 arXiv

  3. [11]

    Recurrent feature reasoning for image inpainting

    Jingyuan Li, Ning Wang, Lefei Zhang, Bo Du, and Dacheng Tao. Recurrent feature reasoning for image inpainting. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7760– 7768, 2020. 2

    2020

  4. [12]

    Mat: Mask-aware transformer for large hole image inpainting

    Wenbo Li, Zhe Lin, Kun Zhou, Lu Qi, Yi Wang, and Jiaya Jia. Mat: Mask-aware transformer for large hole image inpainting. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10758– 10768, 2022. 2

    2022

  5. [13]

    Generalized deep 3d shape prior via part-discretized diffusion pro- cess

    Yuhan Li, Yishun Dou, Xuanhong Chen, Bingbing Ni, Yilin Sun, Yutian Liu, and Fuzhen Wang. Generalized deep 3d shape prior via part-discretized diffusion pro- cess. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 16784–16794, 2023. 1

    2023

  6. [14]

    Megadepth: Learning single-view depth prediction from internet photos

    Zhengqi Li and Noah Snavely. Megadepth: Learning single-view depth prediction from internet photos. In IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), 2018. 4

    2018

  7. [15]

    Image in- painting for irregular holes using partial convolutions

    Guilin Liu, Fitsum A Reda, Kevin J Shih, Ting-Chun Wang, Andrew Tao, and Bryan Catanzaro. Image in- painting for irregular holes using partial convolutions. In Eur. Conf. Comput. Vis., pages 85–100, 2018. 1, 2

    2018

  8. [16]

    Par- tial convolution for padding, inpainting, and image synthesis

    Guilin Liu, Aysegul Dundar, Kevin J Shih, Ting-Chun Wang, Fitsum A Reda, Karan Sapra, Zhiding Yu, Xi- aodong Yang, Andrew Tao, and Bryan Catanzaro. Par- tial convolution for padding, inpainting, and image synthesis. IEEE Transactions on Pattern Analysis and Machine Intelligenc...

    2022

  9. [17]

    Repaint: Inpainting using denoising diffusion probabilistic mod- els

    Andreas Lugmayr, Martin Danelljan, Andrés Romero, Fisher Yu, Radu Timofte, and Luc Van Gool. Repaint: Inpainting using denoising diffusion probabilistic mod- els. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 11451–11461, 2022. 1, 2, 3, 4, 6, 7

    2022

  10. [18]

    Improved denoising diffusion probabilistic models

    Alex Nichol and Prafulla Dhariwal. Improved denoising diffusion probabilistic models. arXiv preprint:2102.09672, 2021. 1, 2, 3

    2021 arXiv

  11. [19]

    Glide: Towards photorealistic image generation and editing with text-guided diffusion mod- els

    Alex Nichol, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob McGrew, Ilya Sutskever, and Mark Chen. Glide: Towards photorealistic image generation and editing with text-guided diffusion mod- els. arXiv preprint:2112.10741, 2021. 3

    2021 arXiv

  12. [20]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fer- nandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Wojciech Galuba, Russell Howes, Po-Yao Huang, Shang-Wen Li, Ishan Misra, Michael Rab...

    2024 arXiv

  13. [21]

    Context encoders: Feature learning by inpainting

    Deepak Pathak, Philipp Krahenbuhl, Jeff Donahue, Trevor Darrell, and Alexei A Efros. Context encoders: Feature learning by inpainting. InIEEE Conf. Comput. Vis. Pattern Recog., pages 2536–2544, 2016. 1, 2

    2016

  14. [22]

    Scalablediffusionmod- els with transformers

    WilliamPeeblesandSainingXie. Scalablediffusionmod- els with transformers. InIEEE/CVF International Con- ference on Computer Vision (ICCV), pages 4172–4182. IEEE, 2023. 3

    2023

  15. [23]

    Habitat-matterport 3d dataset (HM3d): 1000 large-scale 3d environments for embod- ied AI

    Santhosh Kumar Ramakrishnan, Aaron Gokaslan, Erik Wijmans, Oleksandr Maksymets, Alexander Clegg, John M Turner, Eric Undersander, Wojciech Galuba, Andrew Westbury, Angel X Chang, Manolis Savva, Yili Zhao, and Dhruv Batra. Habitat-matterport 3d dataset (HM3d): 1000 large-scale ...

    2021

  16. [24]

    Lee, Jonathan Ho, Tim Salimans, David J

    Chitwan Saharia, William Chan, Huiwen Chang, Chris A. Lee, Jonathan Ho, Tim Salimans, David J. Fleet, and Mohammad Norouzi. Palette: Image-to-image dif- fusion models. InSIGGRAPH, pages 15:1–15:10. ACM,

  17. [25]

    Structure-from-motion revisited

    Johannes Lutz Schönberger and Jan-Michael Frahm. Structure-from-motion revisited. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 4

    2016

  18. [26]

    Weiss, Niru Mah- eswaranathan, and Surya Ganguli

    Jascha Sohl-Dickstein, Eric A. Weiss, Niru Mah- eswaranathan, and Surya Ganguli. Deep unsuper- vised learning using nonequilibrium thermodynamics. arXiv:1503.03585, 2015. 1, 2

    2015 arXiv

  19. [27]

    Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole

    Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score- based generative modeling through stochastic differen- tial equations.arXiv:2011.13456, 2020. 2 9 3D-Consistent Image Inpainting with Diffusion Models

    2011 arXiv

  20. [28]

    Weinzaepfel, V

    P. Weinzaepfel, V. Leroy, T. Lucas, R. Brégier, Y. Cabon, V. Arora, L. Antsfeld, B. Chidlovskii, G. Csurka, and J. Revaud. CroCo: Self-Supervised Pretraining for 3D Vision Tasks by Cross-View Completion. InNeurIPS,

  21. [29]

    Croco v2: Improved cross-view completion pre-training for stereo matching and optical flow

    Philippe Weinzaepfel, Thomas Lucas, Vincent Leroy, Yohann Cabon, Vaibhav Arora, Romain Brégier, Gabriela Csurka, Leonid Antsfeld, Boris Chidlovskii, and Jérôme Revaud. Croco v2: Improved cross-view completion pre-training for stereo matching and optical flow. In European Confe...

    2023

  22. [30]

    Zamir, Zhiyang He, Alexander Sax, Jitendra Malik, and Silvio Savarese

    Fei Xia, Amir R. Zamir, Zhiyang He, Alexander Sax, Jitendra Malik, and Silvio Savarese. Gibson env: Real- world perception for embodied agents. InIEEE Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 9068–9079, 2018. 5

    2018

  23. [31]

    Deep learning for image inpainting: A survey.Pattern Recognition, 134: 109046, 2023

    Hanyu Xiang, Qin Zou, Muhammad Ali Nawaz, Xian- feng Huang, Fan Zhang, and Hongkai Yu. Deep learning for image inpainting: A survey.Pattern Recognition, 134: 109046, 2023. 5

    2023

  24. [32]

    Diverse inpainting and editing with gan inversion.arXiv preprint:2307.15033,

    Ahmet Burak Yildirim, Hamza Pehlivan, Bahri Batuhan Bilecen, and Aysegul Dundar. Diverse inpainting and editing with gan inversion.arXiv preprint:2307.15033,

    arXiv

  25. [33]

    Free-form image inpainting with gated convolution

    Jiahui Yu, Zhe Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S Huang. Free-form image inpainting with gated convolution. InProceedings of the IEEE/CVF In- ternational Conference on Computer Vision, pages 4471– 4480, 2019. 2

    2019

  26. [34]

    Inpaint anything: Seg- ment anything meets image inpainting.arXiv preprint arXiv:2304.06790, 2023

    TaoYu,RunsengFeng,RuoyuFeng,JinmingLiu,XinJin, Wenjun Zeng, and Zhibo Chen. Inpaint anything: Seg- ment anything meets image inpainting.arXiv preprint arXiv:2304.06790, 2023. 3

    2023 arXiv

  27. [35]

    High-fidelity image inpainting with gan inversion

    Yongsheng Yu, Libo Zhang, Heng Fan, and Tiejian Luo. High-fidelity image inpainting with gan inversion. In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XVI, pages 242–258. Springer, 2022. 2

    2022

  28. [36]

    Ageneralprotocoltoprobelargevisionmod- els for 3d physical understanding.arXiv:2310.06836,

    Guanqi Zhan, Chuanxia Zheng, Weidi Xie, and Andrew Zisserman. Ageneralprotocoltoprobelargevisionmod- els for 3d physical understanding.arXiv:2310.06836,

    arXiv

  29. [37]

    Jaakkola, and Shiyu Chang

    Guanhua Zhang, Jiabao Ji, Yang Zhang, Mo Yu, Tommi S. Jaakkola, and Shiyu Chang. Towards coher- ent image inpainting using denoising diffusion implicit models. InICML, pages 41164–41193. PMLR, 2023. 2

    2023

  30. [38]

    Efros, Eli Shecht- man, and Oliver Wang

    Richard Zhang, Phillip Isola, Alexei A. Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric, 2018. 5

    2018

  31. [39]

    Large scale im- age completion via co-modulated generative adversarial networks

    Shengyu Zhao, Jonathan Cui, Yilun Sheng, Yue Dong, Xiao Liang, Eric I Chang, and Yan Xu. Large scale im- age completion via co-modulated generative adversarial networks. arXiv preprint:2103.10428, 2021. 2

    2021 arXiv

  32. [40]

    Generalized decoding for pixel, image, and language.arXiv preprint:2212.11270, 2022

    Xueyan Zou, Zi-Yi Dou, Jianwei Yang, Zhe Gan, Linjie Li, Chunyuan Li, Xiyang Dai, Harkirat Behl, Jianfeng Wang, Lu Yuan, et al. Generalized decoding for pixel, image, and language.arXiv preprint:2212.11270, 2022. 2 10

    2022 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.