Pith. sign in

REVIEW 4 major objections 6 minor 48 references

ORIDa: Object-centric Real-world Image Composition Dataset

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read ORIDa introduces a 30,000-image real-captured dataset that makes object removal and insertion trainable without synthetic data, with models fine-tuned on it beating prior methods in user studies.

desk verdict A genuinely useful real-captured compositing dataset with careful collection, but the evaluation overreaches and the counterfactual alignment assumption is unquantified. read the letter →

arxiv 2506.08964 v1 pith:RCEXJ2RF submitted 2025-06-10 cs.CV

Jinwoo Kim , Sangmin Han , Jinho Jeong , Jiwoo Choi , Dongyoung Kim , Seon Joo Kim This is my paper
classification cs.CV
keywords objectcompositingimagecompositiondatasetfactual-counterfactualpairsremovalinsertiondiffusionmodelfine-tuningreal-worldcaptureRAWdata
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

ORIDa is a proposed dataset for object compositing: placing an object into an image so it looks genuinely present. The paper claims it is the first large-scale, real-captured, publicly available dataset of this kind, with over 30,000 images of 200 objects, each seen in roughly 50 scenes and in up to four positions per scene. The central idea is that paired factual-counterfactual captures, the same scene with and without an object, let a model learn the full visual effect of an object, including shadows and reflections, without synthetic data. The authors show that a standard inpainting diffusion model fine-tuned on ORIDa alone outperforms prior methods for object removal and insertion in user studies and automatic metrics.

What carries the argument

The factual-counterfactual (F-CF) capture protocol is the load-bearing mechanism. Each F-CF set consists of five consecutive images of one scene from a tripod: one background-only counterfactual and four factual images with the object in different positions, with shutter speed, ISO, white balance, and focus fixed. The background-only image serves as ground truth for object removal and as the target condition for insertion, and the four positions create training signal for repositioning and for learning object-to-scene effects such as shadows and reflections.

What would settle it

Compare each background-only image with the object-present images in regions away from the object: if pixel differences outside the object mask are consistently nonzero and grow with capture time, the counterfactual assumption fails. A direct version of this test is to place a static fiducial marker in each scene and measure whether its appearance drifts across the five captures.

Watch

Extended reading notes

Core claim

The paper's central claim is that a real-captured dataset is sufficient to train photorealistic object compositing, provided the data are structured as factual-counterfactual sets: for each scene, one background-only image and four images with the object present, captured consecutively with a tripod and fixed camera settings. These sets expose object-to-scene effects, such as shadows and reflections, that synthetic compositing data cannot supply, while multiple positions per scene expose scene-to-object effects on the object's appearance. The paper reports that fine-tuning Stable Diffusion Inpainting on ORIDa, with only real COCO images added for the insertion task, beats Copy and Paste, Paint-by-Example, AnyDoor, and ObjectStitch on insertion, and beats SD-Inpaint, LaMa, and MGIE on removal, as measured by user preference and by metrics such as PSNR, DINO, CLIP, and LPIPS.

Load-bearing premise

The background-only image is a true counterfactual: between the background capture and the four object-present captures, nothing changes except the object's presence.

Editorial extensions

If this is right

  • A model trained on ORIDa alone can remove an object and erase its shadows and reflections without a separate harmonization or shadow-removal stage.
  • Object insertion can be trained without synthetic compositing pipelines, with only real images such as COCO needed to support identity preservation.
  • Multiple positions per scene support object repositioning as a task, not just removal or insertion.
  • RAW DNG files plus ISP augmentations allow training across color and lighting variations from the same captured content.

Reading between the lines

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

  • Read within its own stated scope, the dataset was collected for rigid, portable, non-human objects, so the demonstrated gains apply to that class until tested on deformable or living subjects.
  • The same capture protocol could be extended to video or multi-view capture, which would add temporal and geometric consistency cues beyond what still-image pairs provide.
  • A direct test of the dataset's value would be to train the same model on matched-scale synthetic pairs and compare shadow and reflection fidelity, since the paper's comparison to prior methods does not isolate this factor.
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 / 6 minor

Summary. ORIDa is a new large-scale, real-captured dataset for object compositing, containing over 30,000 images of 200 objects. Each object appears in an average of roughly 50 scenes, organized into factual-counterfactual (F-CF) sets, where one background-only image accompanies four images of the object in different positions, and factual-only (F-Only) images that provide additional scene diversity. The dataset includes raw DNG files, captions, object points, bounding boxes, and SAM2-generated segmentation masks. The authors also fine-tune Stable Diffusion Inpaint (SD-Inpaint) for object removal and insertion, reporting qualitative results, user studies, and automatic metrics that favor models trained on ORIDa over several baselines. The paper's central claim is that ORIDa is the first publicly available, large-scale, real-captured dataset for object compositing and that it enables improved realism in object removal and insertion without synthetic training data.

Significance. If the core capture assumption holds, ORIDa would be a substantial community resource: it is the first public dataset of this scale to combine real-captured factual-counterfactual pairs, multiple object positions per scene, multiple scenes per object, raw DNG flexibility, and rich annotations. The scale and design are clear advances over ObjectDrop, which is neither public nor multi-scene. The paper also contributes a concrete fine-tuning recipe for SD-Inpaint and reports both user and automatic evaluations. However, the empirical validation currently falls short of fully supporting the dataset's central value claim, primarily because the counterfactual-consistency assumption is asserted rather than quantitatively validated, and because the insertion experiments combine ORIDa with additional COCO data and a modified inference procedure, making the isolated contribution of ORIDa unclear. These issues are fixable through additional validation and control experiments, so the manuscript warrants major revision rather than rejection.

major comments (4)
  1. [§3.2–3.3] The dataset's paired training signal rests on the assumption that each background-only image is a pixel-aligned, photometrically consistent counterfactual of its four factual counterparts. The paper states that camera settings were fixed, tripods and remote controllers were used, and undesirable cases were filtered manually, but it provides no quantitative evidence that residual scene drift, micro-movement, or illumination change is negligible. A tripod-mounted camera can shift slightly when objects are placed or removed, and lighting can drift over the five-shot sequence; manual inspection cannot reliably detect sub-pixel shifts or slow photometric changes. I request quantitative validation: for example, registration residuals between background and factual images in non-object regions, histograms of per-pixel differences under the object mask and outside it, or a comparison against sensor-noise baselines. This is load-bearing because the entire F-CF training signal and the dataset's uniqueness depend on the counterfactual being correct.
  2. [§5.1 and §B.3] The object insertion experiments are not an isolated validation of ORIDa. Training uses an additional 60,000 COCO images with 250,000 object masks, and inference uses a skip-residual modification from DemoFusion that is not part of the pretrained SD-Inpaint pipeline. Consequently, the reported gains in identity preservation, shadow generation, and harmonization could come substantially from the COCO training data, the inference-time skip residual, or their interaction, rather than from ORIDa itself. The paper should ablate these factors, e.g., fine-tune on ORIDa F-CF data alone without COCO, or run the unmodified inference procedure, and report how each component affects the user-study and automatic results. Without such controls, the claim that ORIDa alone supports the observed insertion quality is not established.
  3. [§5.2, Table 2, Table 3, and Figure 11] The user studies are reported without error bars, statistical significance tests, or inter-rater statistics. Table 2 reports mean ratings for 76 participants but no variance or pairwise significance, and Figure 11 reports preference percentages for 62 participants but no confidence intervals or significance tests. The automatic comparison in Table 3 also covers only SD-Inpaint, while the qualitative and user comparisons include LaMa and MGIE; automatic metrics should be reported for all removal baselines. Adding these statistics is necessary to support the strong claims that ORIDa-trained models 'significantly outperform' existing methods.
  4. [§5.2 and §5.3] The evaluation for object removal is conducted on an 'out-held test set' from ORIDa, while insertion is evaluated qualitatively on COCO, internet, and MureCom images. This is reasonable as a start, but the removal numbers on a held-out subset of the same capture campaign may largely reflect the model learning dataset-specific capture conditions rather than generalizable scene understanding. I recommend adding a small cross-dataset or in-the-wild quantitative evaluation for removal (using existing paired or benchmark data where possible), or at least clearly stating and discussing this limitation in the main text.
minor comments (6)
  1. [Throughout] The dataset name is inconsistently spelled as both 'ORIDa' and 'ORIDA' (e.g., in the Introduction and Table 1 row labels); please use a single consistent spelling.
  2. [Figure 1 and Figure 10] There are typographical errors in figure text: 'Factual-Counterfactal' in Figure 1 and 'Obejct' in Figure 10 should be corrected.
  3. [§5.3] The sentence 'adapting its colors seamlessly to the scene and and generating natural shadows' contains a duplicated 'and'.
  4. [§5.2] 'out-held test set' should be 'held-out test set'; please also state how many F-CF sets or images are used for the automatic evaluation in Table 3.
  5. [§3.4 and §5.1] The paper should clarify whether the ISP augmentations (five Lightroom settings) are applied to all images or only to a subset, and how these augmentations interact with the raw DNG files during training; currently this is described only briefly.
  6. [References] Reference [2] is cited as 'MureCom' in the text but the actual title is 'MureObjectStitch'; please align the citation name.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ORIDa's construction and experiments are empirical, and its evaluation is anchored to held-out and external data rather than to fitted parameters.

full rationale

ORIDa is a data-collection paper, not a derivation paper: its deliverable is a physically captured paired dataset (one background-only image plus four object-present images per scene) plus annotations, and its experiments are fine-tuned diffusion models evaluated against a held-out ORIDa test split, COCO images, external MureCom data, and human raters. None of the enumerated circularity patterns applies. (1) Self-definitional: the counterfactual target is captured in the world, not computed from the factual images, so no quantity is defined in terms of a quantity it is claimed to predict or explain. (2) Fitted input called prediction: the only fitted quantities are model weights, and the reported PSNR/DINO/CLIP/LPIPS values and user-study ratings are empirical outcomes of those models, not predictions derived from fitted parameters by construction. (3)-(4) Self-citation and imported uniqueness: the F-CF concept is explicitly credited to ObjectDrop [43], whose authors do not overlap with ORIDa's; no uniqueness theorem is invoked, and no load-bearing argument rests on the authors' own prior work. (5) Ansatz smuggling: the SD-Inpaint backbone, the DemoFusion skip-residual with cosine weighting, SAM2 mask generation, and GPT-4o/Gemini captioning are all external methods cited as such. (6) Renaming: ORIDa contributes new physical captures rather than re-encoding a known empirical pattern in new coordinates. Two legitimate concerns exist but are not circularity: evaluating on a held-out ORIDa split from the same capture campaign may reward dataset-specific artifacts, and the unvalidated counterfactual-capture assumption (tripod stability, fixed lighting across the five-shot sequence) is a data-quality and validity risk. The paper's claims ('first large-scale real-captured public dataset' and 'fine-tuned models outperform baselines in user studies') are contingent empirical statements, not consequences of the dataset's construction by definition.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

This is an empirical dataset paper, so the ledger holds no mathematical free parameters; the listed entries are design choices and assumptions about data validity and evaluation that the central claims rest on.

free parameters (2)
  • number of object positions per F-CF scene = 4
    Hand-chosen design choice: balances coverage of placement effects against capture and annotation effort; no analysis justifies this exact number.
  • number of ISP augmentation settings = 5
    Fixed set of five Adobe Lightroom presets (as-shot, temperature/vibrance variations); chosen arbitrarily and not swept.
assumptions (4)
  • domain assumption The background-only capture is a valid counterfactual: the scene differs from factual images only by object presence.
    Section 3.2 fixes shutter, ISO, WB, focus, tripod and uses remote capture, but subtle lighting or scene changes can remain; manual filtering (Section 3.3) attempts to remove such cases, but no guarantee.
  • domain assumption GPT-4o/Gemini captions and SAM2 segmentation masks are accurate enough for training.
    Section 3.4 relies on these models with only manual point seeds, not full manual verification of outputs.
  • domain assumption Fine-tuning SD-Inpaint is a representative proxy for dataset utility.
    Only one pretrained model family is used; conclusions might not transfer to other generative architectures.
  • domain assumption User study ratings reflect actual composition quality.
    Tables 2 and Figure 11 report preference percentages from 76 and 62 participants without confidence intervals or significance testing.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ORIDa: Object-centric Real-world Image Composition Dataset." pith.science (2026). https://pith.science/paper/RCEXJ2RF

@misc{pith2026250608964,
  author       = {Pith},
  title        = {Pith review of: ORIDa: Object-centric Real-world Image Composition Dataset},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RCEXJ2RF}},
  note         = {Machine review of arXiv:2506.08964}
}
read the original abstract

Object compositing, the task of placing and harmonizing objects in images of diverse visual scenes, has become an important task in computer vision with the rise of generative models. However, existing datasets lack the diversity and scale required to comprehensively explore real-world scenarios. We introduce ORIDa (Object-centric Real-world Image Composition Dataset), a large-scale, real-captured dataset containing over 30,000 images featuring 200 unique objects, each of which is presented across varied positions and scenes. ORIDa has two types of data: factual-counterfactual sets and factual-only scenes. The factual-counterfactual sets consist of four factual images showing an object in different positions within a scene and a single counterfactual (or background) image of the scene without the object, resulting in five images per scene. The factual-only scenes include a single image containing an object in a specific context, expanding the variety of environments. To our knowledge, ORIDa is the first publicly available dataset with its scale and complexity for real-world image composition. Extensive analysis and experiments highlight the value of ORIDa as a resource for advancing further research in object compositing.

Figures

Figures reproduced from arXiv: 2506.08964 by the authors.

Figure 1
Figure 1. Overview of ORIDa. ORIDa contains 200 unique objects and over 30,000 real-captured images, including factual-counterfactual (F-CF) sets and factual-only (F-Only) images. F-CF sets consist of five images: one background-only and four with the object in different positions. F-Only images capture objects in diverse scenes, enhancing the diversity of the dataset for object reposition tasks. Abstract Object compositing, … view at source ↗
Figure 2
Figure 2. Examples of Factual-Counterfactual (F-CF) Sets and Factual-Only (F-Only) Images. The left side shows F-CF sets, consisting of one background-only image and four object-inserted images captured with the object in different positions. The right side displays F-Only images, which feature objects in diverse scenes without corresponding background-only images. vent of diffusion models [12, 29, 31, 36, 37]. For exam￾ple, … view at source ↗
Figure 3
Figure 3. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Annotation examples. Each object includes detailed annotations such as captions, object points, bounding boxes, and segmentation masks. scenes. To ensure consistency within each set, we fixed key camera settings such as shutter speed, ISO, WB, and focus, during a singl…
Figure 6
Figure 6. Figure 6: Dataset statistics per object and attribute. The top chart displays the number of images per object, sorted by y-value, for both factual-only and factual-counterfactual sets. The bottom charts present the percentage distribution of objects based on key attributes: numb…
Figure 7
Figure 7. Figure 7: Indoor/Outdoor ratio. Distribution of indoor and out￾door scenes for factual-counterfactual sets (F-CF), factual-only images (F-Only), and the entire dataset. 4. Dataset Statistics We present several statistics and analyses of our dataset [PITH_FULL_IMAGE:figures/full…
Figure 8
Figure 8. Figure 8: Visualization of color variations across objects. Example objects are shown with their respective color distributions plotted based on red and green channel intensities. Each plot highlights how the appearance of objects varies under different lighting conditions and b…
Figure 9
Figure 9. Figure 9: Object removal - qualitative results across different methods. SD-Inpaint [31], LaMa [40], MGIE [10], and SD-Ours. For MGIE, a text prompt such as “remove the hot sauce from the photo” (for the first row) is used to instruct the model. As shown in [PITH_FULL_IMAGE:fig…
Figure 10
Figure 10. Figure 10: Object insertion - qualitative results. For each row, the Source Object is inserted into the Target Image. Results illustrate differences in identity preservation, shadow generation, color harmonization, and overall realism. “Which one is the best regarding [X]?” [X] …
Figure 11
Figure 11. Figure 11: Object insertion - user study. Participants evaluated different methods on four criteria: object identity preservation, shadow generation (object-to-scene effects), color harmonization (scene-to-object effects), and overall quality based on participants’ preference. C…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 32 canonical work pages

  1. [1]

    Frequency domain image trans- lation: More photo-realistic, better identity-preserving

    Mu Cai, Hong Zhang, Huijuan Huang, Qichuan Geng, Yix- uan Li, and Gao Huang. Frequency domain image trans- lation: More photo-realistic, better identity-preserving. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 13930–13940, 2021. 2

    work page 2021

  2. [2]

    Mureobjectstitch: Multi-reference image composition

    Jiaxuan Chen, Bo Zhang, Qingdong He, Jinlong Peng, and Li Niu. Mureobjectstitch: Multi-reference image composition. arXiv preprint arXiv:2411.07462, 2024. 10

    arXiv 2024

  3. [3]

    Anydoor: Zero-shot object-level im- age customization

    Xi Chen, Lianghua Huang, Yu Liu, Yujun Shen, Deli Zhao, and Hengshuang Zhao. Anydoor: Zero-shot object-level im- age customization. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 6593–6602, 2024. 2, 7, 10

    work page 2024

  4. [4]

    Freecompose: Generic zero- shot image composition with diffusion prior

    Zhekai Chen, Wen Wang, Zhen Yang, Zeqing Yuan, Hao Chen, and Chunhua Shen. Freecompose: Generic zero- shot image composition with diffusion prior. arXiv preprint arXiv:2407.04947, 2024. 2, 3

    arXiv 2024

  5. [5]

    Dovenet: Deep image harmonization via domain verification

    Wenyan Cong, Jianfu Zhang, Li Niu, Liu Liu, Zhixin Ling, Weiyuan Li, and Liqing Zhang. Dovenet: Deep image harmonization via domain verification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8394–8403, 2020. 2

    work page 2020

  6. [6]

    Bargainnet: Background-guided domain translation for image harmonization

    Wenyan Cong, Li Niu, Jianfu Zhang, Jing Liang, and Liqing Zhang. Bargainnet: Background-guided domain translation for image harmonization. In 2021 IEEE International Con- ference on Multimedia and Expo (ICME), pages 1–6. IEEE,

    work page 2021

  7. [7]

    Gemini pro

    Google DeepMind. Gemini pro. https://deepmind. google/technologies/gemini/pro/ , 2024. Ac- cessed: 2024-10-24. 5

    work page 2024

  8. [8]

    Demofusion: Democratising high- resolution image generation with no $$$

    Ruoyi Du, Dongliang Chang, Timothy Hospedales, Yi-Zhe Song, and Zhanyu Ma. Demofusion: Democratising high- resolution image generation with no $$$. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6159–6168, 2024. 9

    work page 2024

Show all 48 references
  1. [9]

    Inpainting with diffusers

    Hugging Face. Inpainting with diffusers. https : / / huggingface . co / docs / diffusers / using - diffusers/inpaint, 2024. Accessed: 2024-11-13. 6, 9, 10

    2024

  2. [10]

    Guiding instruction-based im- age editing via multimodal large language models

    Tsu-Jui Fu, Wenze Hu, Xianzhi Du, William Yang Wang, Yinfei Yang, and Zhe Gan. Guiding instruction-based im- age editing via multimodal large language models. arXiv preprint arXiv:2309.17102, 2023. 6, 7, 10

    2023 arXiv

  3. [11]

    Pct-net: Full resolution image harmonization using pixel-wise color transformations

    Julian Jorge Andrade Guerreiro, Mitsuru Nakazawa, and Bj¨orn Stenger. Pct-net: Full resolution image harmonization using pixel-wise color transformations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5917–5926, 2023. 2

    2023

  4. [12]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. Advances in Neural Information Processing Systems, 33:6840–6851, 2020. 3

    2020

  5. [13]

    Shadow generation for composite image in real-world scenes

    Yan Hong, Li Niu, and Jianfu Zhang. Shadow generation for composite image in real-world scenes. In Proceedings of the AAAI conference on Artificial Intelligence , pages 914–922,

  6. [14]

    Mask-shadowgan: Learning to remove shadows from unpaired data

    Xiaowei Hu, Yitong Jiang, Chi-Wing Fu, and Pheng-Ann Heng. Mask-shadowgan: Learning to remove shadows from unpaired data. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision , pages 2472–2481,

  7. [15]

    Ssh: A self-supervised framework for image harmonization

    Yifan Jiang, He Zhang, Jianming Zhang, Yilin Wang, Zhe Lin, Kalyan Sunkavalli, Simon Chen, Sohrab Amirghodsi, Sarah Kong, and Zhangyang Wang. Ssh: A self-supervised framework for image harmonization. In Proceedings of the IEEE/CVF International Conference on Computer Vision , ...

    2021

  8. [16]

    Harmonizer: Learning to perform white-box image and video harmonization

    Zhanghan Ke, Chunyi Sun, Lei Zhu, Ke Xu, and Rynson WH Lau. Harmonizer: Learning to perform white-box image and video harmonization. In European Conference on Computer Vision, pages 690–706. Springer, 2022. 2

    2022

  9. [17]

    Adam: A method for stochastic opti- mization

    Diederik P Kingma. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980, 2014. 9

    2014 arXiv

  10. [18]

    Geometry-preserved im- age editing

    Taeeun Kwon and Junseok Kwon. Geometry-preserved im- age editing. Electronics Letters, 60(17):e70011, 2024. 2

    2024

  11. [19]

    From shadow segmentation to shadow removal

    Hieu Le and Dimitris Samaras. From shadow segmentation to shadow removal. In European Conference on Computer Vision, pages 264–281. Springer, 2020. 2

    2020

  12. [20]

    Counterfactuals

    David Lewis. Counterfactuals. John Wiley & Sons, 2013. 2

    2013

  13. [21]

    Dreamedit: Subject-driven image editing

    Tianle Li, Max Ku, Cong Wei, and Wenhu Chen. Dreamedit: Subject-driven image editing. arXiv preprint arXiv:2306.12624, 2023. 2, 3

    2023 arXiv

  14. [22]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In European Conference on Computer Vision, pages 740–755. Springer, 2014. 6

    2014

  15. [23]

    Arshadowgan: Shadow generative adversarial network for augmented reality in sin- gle light scenes

    Daquan Liu, Chengjiang Long, Hongpan Zhang, Hanning Yu, Xinzhi Dong, and Chunxia Xiao. Arshadowgan: Shadow generative adversarial network for augmented reality in sin- gle light scenes. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition , pa...

    2020

  16. [24]

    From shadow generation to shadow removal

    Zhihao Liu, Hui Yin, Xinyi Wu, Zhenyao Wu, Yang Mi, and Song Wang. From shadow generation to shadow removal. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4927–4936, 2021. 2

    2021

  17. [25]

    Sgdr: Stochas- tic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. Sgdr: Stochas- tic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983, 2016. 9

    2016 arXiv

  18. [26]

    Dream- com: Finetuning text-guided inpainting model for image composition

    Lingxiao Lu, Jiangtong Li, Bo Zhang, and Li Niu. Dream- com: Finetuning text-guided inpainting model for image composition. arXiv preprint arXiv:2309.15508, 2023. 2

    2023 arXiv

  19. [27]

    Tf-icon: Diffusion-based training-free cross-domain image composi- tion

    Shilin Lu, Yanzhu Liu, and Adams Wai-Kin Kong. Tf-icon: Diffusion-based training-free cross-domain image composi- tion. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, pages 2294–2305, 2023. 2, 3

    2023

  20. [28]

    Hello gpt-4o

    OpenAI. Hello gpt-4o. https : / / openai . com / index/hello-gpt-4o/ , 2024. Accessed: 2024-10-24. 5

    2024

  21. [29]

    Sdxl: Improving latent diffusion models for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. Sdxl: Improving latent diffusion models for high-resolution image synthesis. In International Con- ference on Learning Representations, 2023. 3

    2023

  22. [30]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos. arXiv preprint arXiv:2408.00714, 2024. 5

    2024 arXiv

  23. [31]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10684–10695, 2022. 3, 6, 7, 9, 10

    2022

  24. [32]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Inter- vention, pages 234–241. Springer, 2015. 6

    2015

  25. [33]

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

    Nataniel Ruiz, Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman. Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 22500...

    2023

  26. [34]

    Geodiffuser: Geometry-based image edit- ing with diffusion models

    Rahul Sajnani, Jeroen Vanbaar, Jie Min, Kapil Katyal, and Srinath Sridhar. Geodiffuser: Geometry-based image edit- ing with diffusion models. arXiv preprint arXiv:2404.14403,

    arXiv

  27. [35]

    Collage diffusion

    Vishnu Sarukkai, Linden Li, Arden Ma, Christopher R ´e, and Kayvon Fatahalian. Collage diffusion. In Proceedings of the IEEE/CVF Winter conference on Applications of Computer Vision, pages 4208–4217, 2024. 2

    2024

  28. [36]

    Denois- ing diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models. In International Conference on Learning Representations, 2021. 3

    2021

  29. [37]

    Score-based generative modeling through stochastic differential equa- tions

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Ab- hishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equa- tions. In International Conference on Learning Represen- tations, 2021. 3

    2021

  30. [38]

    Object- stitch: Object compositing with diffusion model

    Yizhi Song, Zhifei Zhang, Zhe Lin, Scott Cohen, Brian Price, Jianming Zhang, Soo Ye Kim, and Daniel Aliaga. Object- stitch: Object compositing with diffusion model. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18310–18319, 2023....

    2023

  31. [39]

    Imprint: Generative object compositing by learning identity-preserving representation

    Yizhi Song, Zhifei Zhang, Zhe Lin, Scott Cohen, Brian Price, Jianming Zhang, Soo Ye Kim, He Zhang, Wei Xiong, and Daniel Aliaga. Imprint: Generative object compositing by learning identity-preserving representation. In Proceedings of the IEEE/CVF Conference on Computer Vision ...

    2024

  32. [40]

    Resolution-robust large mask inpainting with fourier convolutions

    Roman Suvorov, Elizaveta Logacheva, Anton Mashikhin, Anastasia Remizova, Arsenii Ashukha, Aleksei Silvestrov, Naejin Kong, Harshith Goka, Kiwoong Park, and Victor Lempitsky. Resolution-robust large mask inpainting with fourier convolutions. In Proceedings of the IEEE/CVF Win- ...

    2022

  33. [41]

    Stacked condi- tional generative adversarial networks for jointly learning shadow detection and shadow removal

    Jifeng Wang, Xiang Li, and Jian Yang. Stacked condi- tional generative adversarial networks for jointly learning shadow detection and shadow removal. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1788–1797, 2018. 2

    2018

  34. [42]

    Diffusion models are geometry critics: Single image 3d editing using pre-trained diffusion priors

    Ruicheng Wang, Jianfeng Xiang, Jiaolong Yang, and Xin Tong. Diffusion models are geometry critics: Single image 3d editing using pre-trained diffusion priors. arXiv preprint arXiv:2403.11503, 2024. 2

    2024 arXiv

  35. [43]

    Objectdrop: Bootstrap- ping counterfactuals for photorealistic object removal and in- sertion

    Daniel Winter, Matan Cohen, Shlomi Fruchter, Yael Pritch, Alex Rav-Acha, and Yedid Hoshen. Objectdrop: Bootstrap- ping counterfactuals for photorealistic object removal and in- sertion. arXiv preprint arXiv:2403.18818, 2024. 2, 3, 4, 6, 9

    2024 arXiv

  36. [44]

    Infinite-id: Identity-preserved personalization via id-semantics decoupling paradigm

    Yi Wu, Ziqiang Li, Heliang Zheng, Chaoyue Wang, and Bin Li. Infinite-id: Identity-preserved personalization via id-semantics decoupling paradigm. arXiv preprint arXiv:2403.11781, 2024. 2

    2024 arXiv

  37. [45]

    Dccf: Deep comprehensible color fil- ter learning framework for high-resolution image harmoniza- tion

    Ben Xue, Shenghui Ran, Quan Chen, Rongfei Jia, Binqiang Zhao, and Xing Tang. Dccf: Deep comprehensible color fil- ter learning framework for high-resolution image harmoniza- tion. In European Conference on Computer Vision , pages 300–316. Springer, 2022. 2

    2022

  38. [46]

    Paint by example: Exemplar-based image editing with diffusion mod- els

    Binxin Yang, Shuyang Gu, Bo Zhang, Ting Zhang, Xuejin Chen, Xiaoyan Sun, Dong Chen, and Fang Wen. Paint by example: Exemplar-based image editing with diffusion mod- els. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 18381–18391,

  39. [47]

    Image sculpting: Precise ob- ject editing with 3d geometry control

    Jiraphon Yenphraphai, Xichen Pan, Sainan Liu, Daniele Panozzo, and Saining Xie. Image sculpting: Precise ob- ject editing with 3d geometry control. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4241–4251, 2024. 2

    2024

  40. [48]

    Controlcom: Controllable image composition using diffusion model

    Bo Zhang, Yuxuan Duan, Jun Lan, Yan Hong, Huijia Zhu, Weiqiang Wang, and Li Niu. Controlcom: Controllable image composition using diffusion model. arXiv preprint arXiv:2308.10040, 2023. 2, 3, 6

    2023 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.