Pith. sign in

REVIEW 4 major objections 9 minor 1 cited by

HOComp: Interaction-Aware Human-Object Composition

T0 review · 4 major / 9 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read HOComp composes a foreground object onto a human photo with a natural interaction while keeping object and person appearance unchanged, and reports the best scores on its HOIBench benchmark.

desk verdict Solid new task and benchmark for interaction-aware composition, with a real audit gap: the background-consistency numbers are defined relative to the same FLUX.1 Fill inpaintings used to build the training masks. read the letter →

arxiv 2507.16813 v1 pith:2ZCBKHNC submitted 2025-07-22 cs.CV

Dong Liang , Jinyuan Jia , Yuhao Liu , Rynson W.H. Lau This is my paper
classification cs.CV
keywords human-objectcompositionimagediffusiontransformermultimodalLLMposeguidanceappearancepreservationinteractiondatasetHOIBenchbenchmark
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

HOComp is a diffusion-transformer framework for composing a foreground object, such as a perfume bottle, onto a human-centric background image so that the person actually interacts with it, while both the person's identity and the object's shape and texture remain unchanged. The paper argues that existing composition methods fail at this task because they treat the human as part of the background and only harmonize the inserted object, producing implausible gestures or altering the person. HOComp instead separates the problem into two mechanisms: a coarse-to-fine pose controller that uses a multimodal LLM to decide where and how the interaction happens and supervises pose keypoints only in that region, and an appearance-preservation module combining attention modulation, multi-view comparison, and a background-consistency loss. To train and test it, the paper introduces IHOC, a dataset of 11,700 paired before/during interaction images covering 117 interaction types, and reports that HOComp outperforms nine existing methods on FID, CLIP-Score, HOI-Score, DINO-Score, and SSIM(BG). If these results hold, the method offers a route to automatic, identity-preserving product-in-photo composition and downstream video generation.

What carries the argument

The load-bearing machinery is the pair of modules MRPG and DCAP acting on a fine-tuned FLUX.1 diffusion transformer. MRPG is the coarse-to-fine interaction controller: at the coarse level, the multimodal LLM outputs the text prompt, object bounding box, and interaction region $B_r$, which is converted to a mask and fed as a condition; at the fine level, a pose estimator extracts keypoints from the predicted and ground-truth images and the localized loss $L_p$ supervises only the keypoints inside $B_r$. DCAP is the three-part appearance keeper: shape-aware attention modulation applies a residual adjustment scaled by the object shape prior $M_{shape}$ to the foreground-relevant attention maps, the multi-view appearance loss uses a multi-view generator and CLIP features to enforce semantic texture agreement, and the background consistency loss $L_{background}$ computes L2 distance between predicted and ground-truth pixels where the unchanged-region mask $M_{unchanged}$ is active. The full training objective sums the denoising loss with these three terms, and the IHOC dataset supplies the paired images and masks needed to supervise them.

What would settle it

Collect real photographs of the same person immediately before and during an interaction, run FLUX.1 Fill on the during-photo with the object masked out, and compare the inpainted before-image with the real before-image: if pose keypoints, SSIM over the supposed unchanged region, or per-pixel colors differ systematically beyond typical inpainting error, then the training targets and the SSIM(BG) metric encode artifacts of the inpainting model rather than true pre-interaction states.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that interaction-aware human-object composition is best solved by explicitly decoupling what interaction and where it occurs from how the person and object look. Concretely, HOComp's MRPG module asks a multimodal LLM to produce a text prompt, an object box, and an interaction region, then adds a pose-guided loss $L_p$ that measures the distance between predicted and ground-truth pose keypoints only inside that region, forcing the generated pose to reflect the interaction without changing the body elsewhere. The DCAP module then preserves appearance with three complementary signals: a shape-aware attention modulation that nudges foreground-related attention maps toward the object's mask, a multi-view appearance loss that compares CLIP features of synthesized views of the generated object with views of the input object, and a background consistency loss over the unchanged-region mask. The paper reports that on its new HOIBench benchmark, HOComp achieves an FID of 9.27, HOI-Score of 87.39, DINO-Score of 78.21, and SSIM(BG) of 96.57, all the best among the nine compared methods, and ranks first in the user study on image quality, interaction harmonization, and appearance preservation.

Load-bearing premise

The weakest assumption is that the synthetic before-interaction images, made by inpainting the object out of each composited training image, show what the person and scene truly looked like before the interaction, with every interaction-induced change such as shadows, reflections, or cloth deformation captured by the unchanged-region mask.

Editorial extensions

If this is right

  • If the reported scores hold, HOComp is the first method for this task that can take only a human image and an object image and automatically decide both where and how to place the object, removing the manual target-region and text-prompt specification required by prior methods.
  • The method's success suggests that explicitly separating the interaction constraint (pose within a region) from the appearance constraints (object shape and texture, background identity) is enough to avoid the two failure modes of prior composition methods: unnatural gestures and identity drift.
  • Because HOComp can be plugged into an image-to-video model, the same interaction-aware composition could generate product-demonstration videos in which a person naturally holds, lifts, or uses the product.
  • The IHOC dataset, with paired before/after images, interaction regions, and unchanged-region masks across 117 interaction types, provides a training and evaluation resource that did not previously exist for this task.

Reading between the lines

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

  • Beyond the paper: the large HOI-Score gap between HOComp (87.39) and GPT-4o (75.22) suggests that the pose-guided loss, not just the MLLM's text and box proposals, is doing much of the interaction work; a clean test would be to feed the same model oracle interaction regions from the ground-truth masks and see whether the score changes.
  • Beyond the paper: because the benchmark's ground truth for background preservation is itself defined by the same inpainting model used to create the training data, the SSIM(BG) metric may reward models that reproduce the inpainter's style rather than true photographic consistency.
  • Beyond the paper: the same coarse-to-fine pattern could be applied to multi-object scenes, with one interaction region per object, and to video by adding a temporal smoothness term on the pose keypoints; neither is tested in the paper.
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 / 9 minor

Summary. The paper proposes HOComp, a diffusion-transformer-based method for compositing a foreground object onto a human-centric background image while aiming for natural human-object interaction and appearance consistency. The method uses an MLLM (GPT-4o) to generate the interaction prompt, object box, and interaction region; a pose-guided loss within the interaction region; shape-aware attention modulation; a multi-view appearance loss; and a background consistency loss. The authors introduce the IHOC dataset (built from HICO-DET real images and FLUX.1 synthetic images, with backgrounds produced by FLUX.1 Fill inpainting) and the HOIBench benchmark, and report results on FID, CLIP-Score, HOI-Score, DINO-Score, SSIM(BG), and a user study, claiming state-of-the-art performance.

Significance. If the claims hold, HOComp would be a useful advance in interaction-aware composition, a task that existing composition methods handle poorly. The MLLM-driven coarse-to-fine pose guidance is a sensible way to inject interaction semantics, and the combination of attention modulation, multi-view supervision, and background consistency is principled. The paper includes component ablations, a user study, and an extensive appendix. However, the central empirical claims rest on two currently fragile pillars: the background ground truth is produced by an inpainting model whose fidelity is never validated, and the main quantitative benchmark is the authors' own, with hyperparameters selected on that same benchmark and no independent validation split. The dataset and benchmark are not released, which further limits verification. The significance is therefore conditional on addressing the validity and independence concerns.

major comments (4)
  1. [Sec. 3.4, Appendix B.1, Eq. (6), SSIM(BG)] The background human images and unchanged region masks are produced by FLUX.1 Fill inpainting with manually annotated masks, and Appendix B.1 states that the inpainter 'reconstruct[s] plausible human poses without interactions,' explicitly allowing pose changes. The paper never validates that these inpaintings correspond to true pre-interaction states, nor that the manual masks cover all interaction-induced changes such as shadows, reflections, or cloth deformation. Because both the background consistency loss Lbackground (Eq. 6) and the SSIM(BG) metric are computed relative to these same synthetic masks, the reported background-consistency advantage (SSIM(BG)=96.57 vs. AnyDoor 90.71; the large ablation drop to 58.49 without Lb) may largely reward agreement with FLUX.1 Fill's reconstruction rather than faithfulness to the actual pre-interaction scene. The authors should validate inpainting fidelity using real before/after pairs or a human study, and should report background consistency on manually verified unchanged regions.
  2. [Sec. 4 (Benchmark), Appendix Tables 4 and 8] The loss coefficients (alpha_1, alpha_2, alpha_3), the attention modulation strength alpha, and the guidance scale are all selected by sweeping HOIBench, which is the same benchmark on which the final quantitative claims are made. There is no held-out validation split, and baselines are not given the same tuning opportunity. This selection-on-the-test-benchmark protocol inflates the reported FID, CLIP, HOI, DINO, and SSIM(BG) numbers and makes the comparison with the nine (and additional five) baselines difficult to interpret. The authors should fix hyperparameters on a validation split, report the selection procedure, or give all baselines the same tuning budget; in all cases they should report variance or error bars across the 600 evaluation instances.
  3. [Sec. 3.3, Eq. (3), Fig. 2] The shape-aware attention modulation requires a shape prior Mshape, and the text says this mask is 'extracted from the ground-truth image.' At inference there is no ground-truth image, and the paper does not specify how Mshape is obtained at test time. If Mshape is predicted from the input foreground object, the segmentation module must be described and evaluated; if the ground-truth mask is used, the method is not applicable at inference. This gap is load-bearing for the DINO-Score claim (78.21 with SAAM vs. 66.52 without) and for the entire method's practical validity.
  4. [Sec. 4.1, Table 1 (referred to as Table 10)] The comparison protocol states that 'all methods with public training code are retrained or fine-tuned on our dataset.' This is a non-standard protocol for zero-shot baselines such as AnyDoor and Paint by Example, and the paper does not report the amount of data, number of steps, or hyperparameter tuning applied to each baseline. Fine-tuning all baselines on IHOC can either advantage or disadvantage them relative to their original design, and without a controlled protocol the quantitative superiority claim is hard to interpret. Additionally, the evaluation details for HOI-Score (e.g., how UPT detections are matched to ground-truth interactions) and FID (number of generated images and reference set) are not given.
minor comments (9)
  1. [Sec. 4.1 and user study] The main text repeatedly refers to 'Table 10' when the main paper contains only Table 1 and Table 2; the referenced table appears to be Table 10 in Appendix H. Please fix the cross-references to the correct table numbers.
  2. [Sec. 4 and Appendix B.1] The HICO-DET citation is inconsistent: it is reference [5] in the main text (Sec. 3.4) but reference [21] in Sec. 4 and Appendix B.1. Use the same reference consistently.
  3. [Sec. 3.4, Stage 4] The sentence 'Interaction regions are computed by extracting the minimal bounding box of the interaction area within the unchanged region mask' is ambiguous; the interaction area should be the complement of the unchanged region mask, not a subset of it.
  4. [Eq. (1)] If no pose keypoints fall inside the interaction region Br, the denominator n in Lp is zero; please state how this case is handled.
  5. [User study] The user study reports average ranks but provides no significance tests, confidence intervals, or inter-rater agreement; the statement that the results 'significantly outperform' other methods is not supported by the reported statistics.
  6. [Dataset and code release] The IHOC dataset and HOIBench benchmark are not released; a link or a release plan should be included so that the results can be reproduced and the benchmark can be used by the community.
  7. [Fig. 2] Figure 2 is dense and the inference-time path for obtaining the shape mask Mshape is not clearly drawn; the caption should explicitly indicate how Mshape is obtained during inference versus training.
  8. [Sec. 3.2] There is a typo: 'lefting' should be 'lifting.'
  9. [Appendix D] The coefficient sensitivity analysis in Table 4 changes one coefficient at a time while keeping others fixed, but the paper does not discuss how the reported best configuration was selected or whether the selected values generalize beyond HOIBench.

Circularity Check

1 steps flagged · score 4.0 of 10

Headline HOIBench numbers are selected by sweeping hyperparameters on the same benchmark; no direct derivation-level circularity otherwise.

  1. fitted input called prediction [Appendix D Table 4, Appendix F.2 Table 6, Appendix F.4 Table 8; main Table 1]
    "As shown in Table. 8, guidance scale = 3.5 achieves the best overall performance (FID = 9.27, CLIP = 30.29, HOI = 87.39, DINO = 78.21, SSIM(BG) = 96.57)."

    The paper's reported 'Ours' scores in Table 1 are the winning rows of hyperparameter sweeps (loss coefficients, LoRA rank, guidance scale) conducted on HOIBench itself. The final configuration is selected by maximizing HOIBench metrics, and the same HOIBench metrics are then presented as the method's predictive performance. This matches the fitted-input-called-prediction pattern: the reported benchmark scores are selected maxima over the evaluation set rather than independent predictions, so the claimed quantitative superiority is partly forced by the selection procedure.

full rationale

HOComp is an empirical systems paper rather than a derivation chain, so most of the claimed contributions are validated by experiments rather than by equations that reduce to their inputs. The main partial circularity is benchmark selection: the headline numbers in Table 1 are obtained by sweeping hyperparameters on the newly introduced HOIBench and then reporting the best HOIBench scores (Appendix D, F.2, F.4). This makes the headline FID, CLIP, HOI, DINO, and SSIM(BG) values selected maxima over the evaluation benchmark, not out-of-sample predictions. Independent grounding partially offsets this: HOI-Score uses the pretrained UPT detector, DINO-Score and CLIP-Score use external pretrained models, and the user study is a separate subjective judgment on held-out internet images. The only self-citation, [37], appears in a Related Works list and is not load-bearing. The unvalidated FLUX.1 Fill inpainting used to define unchanged-region masks (Sec 3.4 and Appendix B.1) is a real validity risk for the background-consistency claim, especially because the paper states the inpainter is used to 'reconstruct plausible human poses without interactions,' but this is a ground-truth fidelity concern rather than a circularity of derivation. No equation in the paper is equivalent to its inputs by construction. Score 4 reflects the benchmark-selection circularity while acknowledging the independent external components.

Assumptions & free parameters 7 free parameters · 7 assumptions · 0 invented entities

The central empirical claim depends on the validity of synthetic dataset construction via FLUX inpainting, MLLM annotations, and pretrained pose, multi-view, and HOI detectors. These are domain assumptions rather than free parameters. The hyperparameters listed are fitted on the authors' own benchmark.

free parameters (7)
  • alpha_1 pose loss weight = 1.0
    Weight of L_p in Eq. (7); chosen via HOIBench sweeps in Appendix D, affects HOI and pose alignment.
  • alpha_2 background loss weight = 0.5
    Weight of L_b in Eq. (7); increasing to 1.0 improves SSIM(BG) but degrades FID, CLIP, HOI, and DINO; default set on the benchmark.
  • alpha_3 multi-view appearance loss weight = 0.8
    Weight of L_appearance in Eq. (7); default from the coefficient study in Appendix D.
  • alpha attention modulation strength = 1.0
    Modulation strength in Eq. (3); larger values destabilize the pretrained attention distribution, and 1.0 was selected on HOIBench.
  • guidance scale = 3.5
    Inference-time classifier-free guidance scale; swept in Appendix F.4 with 3.5 best on HOIBench.
  • multi-view count k = 6
    Number of views in L_appearance; Table 5 shows monotonic improvement up to k=6.
  • LoRA rank = 16
    Rank for LoRA fine-tuning; Table 6 shows rank 16 best across metrics.
assumptions (7)
  • domain assumption HICO-DET's 117 interaction types are a sufficient ontology for human-object composition.
    Dataset and benchmark both sample from these categories; if real-world interactions fall outside this set, generalization is untested. Sec 3.4.
  • domain assumption DWPose keypoints on generated images are reliable enough to supervise pose.
    L_p in Eq. (1) uses keypoints extracted from predicted images; pose estimator errors propagate into supervision. Sec 3.2.
  • domain assumption Zero123+ multi-view images provide meaningful semantic views for CLIP comparison.
    L_appearance in Eq. (5) relies on Zero123+ outputs; if views are unrealistic, the loss does not measure true appearance consistency. Sec 3.3 and Appendix F.1.
  • domain assumption CLIP embeddings capture appearance consistency between foreground views.
    L_appearance uses cosine similarity in CLIP space; CLIP is imperfect for fine-grained texture and logo preservation. Sec 3.3.
  • domain assumption FLUX.1 Fill inpainting yields faithful 'without object' background human images.
    Dataset background images are created by inpainting the composite; inpainting artifacts become training targets. Sec 3.4, Stage 4.
  • domain assumption Unchanged region masks correctly label pixels unaffected by interaction.
    Background loss forces equality with GT outside the mask; mislabeled masks punish legitimate interaction-related changes such as shadows or clothing deformation. Sec 3.4.
  • domain assumption UPT HOI detector score is a valid proxy for interaction correctness on generated images.
    HOI-Score uses UPT pretrained on HICO-DET; detector errors or distribution shift affect the headline metric. Sec 4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HOComp: Interaction-Aware Human-Object Composition." pith.science (2026). https://pith.science/paper/2ZCBKHNC

@misc{pith2026250716813,
  author       = {Pith},
  title        = {Pith review of: HOComp: Interaction-Aware Human-Object Composition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2ZCBKHNC}},
  note         = {Machine review of arXiv:2507.16813}
}
read the original abstract

While existing image-guided composition methods may help insert a foreground object onto a user-specified region of a background image, achieving natural blending inside the region with the rest of the image unchanged, we observe that these existing methods often struggle in synthesizing seamless interaction-aware compositions when the task involves human-object interactions. In this paper, we first propose HOComp, a novel approach for compositing a foreground object onto a human-centric background image, while ensuring harmonious interactions between the foreground object and the background person and their consistent appearances. Our approach includes two key designs: (1) MLLMs-driven Region-based Pose Guidance (MRPG), which utilizes MLLMs to identify the interaction region as well as the interaction type (e.g., holding and lefting) to provide coarse-to-fine constraints to the generated pose for the interaction while incorporating human pose landmarks to track action variations and enforcing fine-grained pose constraints; and (2) Detail-Consistent Appearance Preservation (DCAP), which unifies a shape-aware attention modulation mechanism, a multi-view appearance loss, and a background consistency loss to ensure consistent shapes/textures of the foreground and faithful reproduction of the background human. We then propose the first dataset, named Interaction-aware Human-Object Composition (IHOC), for the task. Experimental results on our dataset show that HOComp effectively generates harmonious human-object interactions with consistent appearances, and outperforms relevant methods qualitatively and quantitatively.

Figures

Figures reproduced from arXiv: 2507.16813 by the authors.

Figure 1
Figure 1. When compositing a foreground object onto a human-centric background image, existing [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Pipeline of HOComp. Our method includes two core modules: MRPG for constraining human-object interaction and DCAP for maintaining appearance consistency. Inference Phase (left): MRPG uses MLLMs to generate a text prompt C, object box Bo and interaction region Br. Among these, Br and C are encoded and, together with the object ID, detail features, and background features, are used to condition the DiT for final compo… view at source ↗
Figure 3
Figure 3. Visualization of attention maps related to the foreground text embeddings AEf c→X and the identity features AEID→X , both exhibiting strong alignment with object shape. Shape-aware Attention Modulation. To enhance shape consistency, we modulate foreground-relevant attention maps in the MM-DiT blocks, encouraging the attention maps to align more precisely with the object’s shape prior. This design is motivated by the… view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison with six top performing SOTA methods from Table 10. The prompts [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visual comparison of the ablation study in Table 2. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: An example failure case of HOComp. The red boxes indicate the interaction regions. In this paper, we have presented HOComp, a framework for interaction-aware human-object composition. It leverages MLLM￾driven region-based pose guidance (MRPG) for constrained human-obje…
Figure 7
Figure 7. Figure 7: Overview of the construction process of our [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]
Figure 8
Figure 8. Figure 8: Visualization of our Interaction-aware Human-Object Composition (IHOC) Dataset. [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]
Figure 9
Figure 9. Figure 9: Statistical analysis of our Interaction-aware Human-Object Composition (IHOC) dataset across six dimensions: (a) human viewpoint, (b) human pose, (c) interaction body part, (d) foreground object size, (e) image style, and (f) background scene type. These statistics dem…
Figure 10
Figure 10. Figure 10: Visual results of ablation study on attention modulation strategies in Table 3. [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: Visual results of ablation study on view numbers used in multi-view appearance loss. [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]
Figure 12
Figure 12. Figure 12: Visual results of ablation study on multi-view generators. [PITH_FULL_IMAGE:figures/full_fig_p022_12.png]
Figure 13
Figure 13. Figure 13: Ablation study on different backbones for foreground ID encoders. [PITH_FULL_IMAGE:figures/full_fig_p023_13.png]
Figure 14
Figure 14. Figure 14: Ablation study on different guidance scales (denoted as gs) during inference. [PITH_FULL_IMAGE:figures/full_fig_p023_14.png]
Figure 15
Figure 15. Figure 15: Quantitative comparison with recent state-of-the-art multi-modality models. The prompts [PITH_FULL_IMAGE:figures/full_fig_p024_15.png]
Figure 16
Figure 16. Figure 16: Additional qualitative comparisons of our [PITH_FULL_IMAGE:figures/full_fig_p025_16.png]
Figure 17
Figure 17. Figure 17: Additional qualitative results of HOComp. Each example includes: (1) Top: the final composited image, (2) Bottom: the input background human and foreground object. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_17.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. HiddenObjects: Scalable Diffusion-Distilled Spatial Priors for Object Placement

    cs.CV 2026-04 unverdicted novelty 7.0 of 10

    A diffusion-based pipeline creates a 27M-annotation dataset of object placements that outperforms human annotations and baselines on image editing tasks, then distills it into a fast model.

Reference graph

Works this paper leans on

100 extracted references · 53 canonical work pages · cited by 1 Pith paper

  1. [1]

    Break- a-scene: Extracting multiple concepts from a single image

    Omri Avrahami, Kfir Aberman, Ohad Fried, Daniel Cohen-Or, and Dani Lischinski. Break- a-scene: Extracting multiple concepts from a single image. In ACM SIGGRAPH Asia, pages 1–12, 2023

    2023

  2. [2]

    Stephen Batifol, Andreas Blattmann, Frederic Boesel, Saksham Consul, Cyril Diagne, Tim Dockhorn, Jack English, Zion English, Patrick Esser, Sumith Kulal, et al. Flux. 1 kontext: Flow matching for in-context image generation and editing in latent space. arXiv e-prints, pages arXiv–2506, 2025

    2025

  3. [3]

    FLUX.1-dev: A 12B Parameter Rectified Flow Transformer for Text- to-Image Generation

    Black Forest Labs. FLUX.1-dev: A 12B Parameter Rectified Flow Transformer for Text- to-Image Generation. https://huggingface.co/spaces/black-forest-labs/FLUX. 1-dev, 2024

    2024

  4. [4]

    FLUX.1-Fill-dev: A 12B Parameter Rectified Flow Transformer for Inpainting and Outpainting

    Black Forest Labs. FLUX.1-Fill-dev: A 12B Parameter Rectified Flow Transformer for Inpainting and Outpainting. https://huggingface.co/black-forest-labs/FLUX. 1-Fill-dev, 2024

    2024

  5. [5]

    Learning to detect human- object interactions

    Yu-Wei Chao, Yunfan Liu, Xieyang Liu, Huayi Zeng, and Jia Deng. Learning to detect human- object interactions. In Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV), pages 381–389, 2018

    2018

  6. [6]

    Vir- tualmodel: Generating object-id-retentive human-object interaction image by diffusion model for e-commerce marketing

    Binghui Chen, Chongyang Zhong, Wangmeng Xiang, Yifeng Geng, and Xuansong Xie. Vir- tualmodel: Generating object-id-retentive human-object interaction image by diffusion model for e-commerce marketing. arXiv:2405.09985, 2024. 10

    arXiv 2024

  7. [7]

    Mureobjectstitch: Multi- reference image composition

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

    arXiv 2024

  8. [8]

    Zero-shot image editing with reference imitation

    Xi Chen, Yutong Feng, Mengting Chen, Yiyang Wang, Shilong Zhang, Yu Liu, Yujun Shen, and Hengshuang Zhao. Zero-shot image editing with reference imitation. In NeurIPS, volume 37, pages 84010–84032, 2024

    2024

Show all 100 references
  1. [9]

    Anydoor: Zero-shot object-level image customization

    Xi Chen, Lianghua Huang, Yu Liu, Yujun Shen, Deli Zhao, and Hengshuang Zhao. Anydoor: Zero-shot object-level image customization. In CVPR, pages 6593–6602, 2024

    2024

  2. [10]

    Unireal: Universal image generation and editing via learning real-world dynamics

    Xi Chen, Zhifei Zhang, He Zhang, Yuqian Zhou, Soo Ye Kim, Qing Liu, Yijun Li, Jianming Zhang, Nanxuan Zhao, Yilin Wang, et al. Unireal: Universal image generation and editing via learning real-world dynamics. arXiv:2412.07774, 2024

    2024 arXiv

  3. [11]

    Freecom- pose: Generic zero-shot image composition with diffusion prior

    Zhekai Chen, Wen Wang, Zhen Yang, Zeqing Yuan, Hao Chen, and Chunhua Shen. Freecom- pose: Generic zero-shot image composition with diffusion prior. In ECCV, pages 70–87. Springer, 2024

    2024

  4. [12]

    Cinema: Coherent multi-subject video generation via mllm-based guidance

    Yufan Deng, Xun Guo, Yizhi Wang, Jacob Zhiyuan Fang, Angtian Wang, Shenghai Yuan, Yiding Yang, Bo Liu, Haibin Huang, and Chongyang Ma. Cinema: Coherent multi-subject video generation via mllm-based guidance. arXiv:2503.10391, 2025

    2025 arXiv

  5. [13]

    Freecustom: Tuning-free customized image generation for multi-concept composition

    Ganggui Ding, Canyu Zhao, Wen Wang, Zhen Yang, Zide Liu, Hao Chen, and Chunhua Shen. Freecustom: Tuning-free customized image generation for multi-concept composition. InCVPR, pages 9089–9098, 2024

    2024

  6. [14]

    Wish you were here: Context-aware human generation

    Oran Gafni and Lior Wolf. Wish you were here: Context-aware human generation. In CVPR, pages 7840–7849, 2020

    2020

  7. [15]

    Mix-of-show: Decentralized low-rank adaptation for multi-concept customization of diffusion models

    Yuchao Gu, Xintao Wang, Jay Zhangjie Wu, Yujun Shi, Yunpeng Chen, Zihan Fan, Wuyou Xiao, Rui Zhao, Shuning Chang, and Weijia Wu. Mix-of-show: Decentralized low-rank adaptation for multi-concept customization of diffusion models. In NeurIPS, volume 36, pages 15890–15902, 2023

    2023

  8. [16]

    Affordance- aware object insertion via mask-aware dual diffusion

    Jixuan He, Wanhua Li, Ye Liu, Junsik Kim, Donglai Wei, and Hanspeter Pfister. Affordance- aware object insertion via mask-aware dual diffusion. arXiv:2412.14462, 2024

    2024 arXiv

  9. [17]

    Clipscore: A reference-free evaluation metric for image captioning

    Jack Hessel, Ari Holtzman, Maxwell Forbes, and Yejin Choi. Clipscore: A reference-free evaluation metric for image captioning. In EMNLP, 2021

    2021

  10. [18]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium. In NeurIPS, 2017

    2017

  11. [19]

    Interactd- iffusion: Interaction control in text-to-image diffusion models

    Jiun Tian Hoe, Xudong Jiang, Chee Seng Chan, Yap-Peng Tan, and Weipeng Hu. Interactd- iffusion: Interaction control in text-to-image diffusion models. In CVPR, pages 6180–6189, 2024

    2024

  12. [20]

    Viewdiff: 3d-consistent image generation with text-to-image models

    Lukas Höllein, Aljaž Boži ˇc, Norman Müller, David Novotny, Hung-Yu Tseng, Christian Richardt, Michael Zollhöfer, and Matthias Nießner. Viewdiff: 3d-consistent image generation with text-to-image models. In CVPR, pages 5043–5052, 2024

    2024

  13. [21]

    Visual compositional learning for human-object interaction detection

    Zhi Hou, Xiaojiang Peng, Yu Qiao, and Dacheng Tao. Visual compositional learning for human-object interaction detection. In ECCV, pages 584–600. Springer, 2020

    2020

  14. [22]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. In ICLR, 2022

    2022

  15. [23]

    Personahoi: Effortlessly improving personalized face with human-object interaction generation

    Xinting Hu, Haoran Wang, Jan Eric Lenssen, and Bernt Schiele. Personahoi: Effortlessly improving personalized face with human-object interaction generation. In CVPR, 2025

    2025

  16. [24]

    Exploiting relationship for complex-scene image generation

    Tianyu Hua, Hongdong Zheng, Yalong Bai, Wei Zhang, Xiao-Ping Zhang, and Tao Mei. Exploiting relationship for complex-scene image generation. In AAAI, volume 35, pages 1584–1592, 2021. 11

    2021

  17. [25]

    Dreamfuse: Adaptive image fusion with diffusion transformer

    Junjia Huang, Pengxiang Yan, Jiyang Liu, Jie Wu, Zhao Wang, Yitong Wang, Liang Lin, and Guanbin Li. Dreamfuse: Adaptive image fusion with diffusion transformer. arXiv:2504.08291, 2025

    2025 arXiv

  18. [26]

    Learning disentangled identifiers for action-customized text-to-image generation

    Siteng Huang, Biao Gong, Yutong Feng, Xi Chen, Yuqian Fu, Yu Liu, and Donglin Wang. Learning disentangled identifiers for action-customized text-to-image generation. In CVPR, pages 7797–7806, 2024

    2024

  19. [27]

    Reversion: Diffusion- based relation inversion from images

    Ziqi Huang, Tianxing Wu, Yuming Jiang, Kelvin CK Chan, and Ziwei Liu. Reversion: Diffusion- based relation inversion from images. In SIGGRAPH Asia, pages 1–11, 2024

    2024

  20. [28]

    Hunyuanvideo-homa: Generic human-object interaction in multimodal driven human animation

    Ziyao Huang, Zixiang Zhou, Juan Cao, Yifeng Ma, Yi Chen, Zejing Rao, Zhiyong Xu, Hongmei Wang, Qin Lin, Yuan Zhou, et al. Hunyuanvideo-homa: Generic human-object interaction in multimodal driven human animation. arXiv preprint arXiv:2506.08797, 2025

    2025 arXiv

  21. [29]

    Record: Reasoning and correcting diffusion for hoi generation

    Jian-Yu Jiang-Lin, Kang-Yang Huang, Ling Lo, Yi-Ning Huang, Terence Lin, Jhih-Ciang Wu, Hong-Han Shuai, and Wen-Huang Cheng. Record: Reasoning and correcting diffusion for hoi generation. In ACM MM, pages 9465–9474, 2024

    2024

  22. [30]

    Dense text-to-image generation with attention modulation

    Yunji Kim, Jiyoung Lee, Jin-Hwa Kim, Jung-Woo Ha, and Jun-Yan Zhu. Dense text-to-image generation with attention modulation. In ICCV, pages 7701–7711, 2023

    2023

  23. [31]

    Auto-encoding variational bayes

    Diederik P Kingma and Max Welling. Auto-encoding variational bayes. arXiv:1312.6114, 2013

    2013 arXiv

  24. [32]

    Omg: Occlusion-friendly personalized multi-concept generation in diffusion models

    Zhe Kong, Yong Zhang, Tianyu Yang, Tao Wang, Kaihao Zhang, Bizhu Wu, Guanying Chen, Wei Liu, and Wenhan Luo. Omg: Occlusion-friendly personalized multi-concept generation in diffusion models. In ECCV, pages 253–270. Springer, 2024

    2024

  25. [33]

    Multi- concept customization of text-to-image diffusion

    Nupur Kumari, Bingliang Zhang, Richard Zhang, Eli Shechtman, and Jun-Yan Zhu. Multi- concept customization of text-to-image diffusion. In CVPR, pages 1931–1941, 2023

    1931

  26. [34]

    Bifrost: 3d-aware image compositing with language instructions

    Lingxiao Li, Kaixiong Gong, Wei-Hong Li, Tao Chen, Xiaojun Yuan, and Xiangyu Yue. Bifrost: 3d-aware image compositing with language instructions. In NeurIPS, volume 37, pages 129480– 129506, 2024

    2024

  27. [35]

    Tuning-free image customization with image and text guidance

    Pengzhi Li, Qiang Nie, Ying Chen, Xi Jiang, Kai Wu, Yuhuan Lin, Yong Liu, Jinlong Peng, Chengjie Wang, and Feng Zheng. Tuning-free image customization with image and text guidance. In ECCV, pages 233–250. Springer, 2024

    2024

  28. [36]

    Gligen: Open-set grounded text-to-image generation

    Yuheng Li, Haotian Liu, Qingyang Wu, Fangzhou Mu, Jianwei Yang, Jianfeng Gao, Chunyuan Li, and Yong Jae Lee. Gligen: Open-set grounded text-to-image generation. In CVPR, pages 22511–22521, 2023

    2023

  29. [37]

    V odiff: Controlling object visibility order in text-to-image generation

    Dong Liang, Jinyuan Jia, Yuhao Liu, Zhanghan Ke, Hongbo Fu, and Rynson WH Lau. V odiff: Controlling object visibility order in text-to-image generation. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 18379–18389, 2025

    2025

  30. [38]

    Movie weaver: Tuning-free multi-concept video personalization with anchored prompts

    Feng Liang, Haoyu Ma, Zecheng He, Tingbo Hou, Ji Hou, Kunpeng Li, Xiaoliang Dai, Felix Juefei-Xu, Samaneh Azadi, Animesh Sinha, et al. Movie weaver: Tuning-free multi-concept video personalization with anchored prompts. arXiv:2502.07802, 2025

    2025 arXiv

  31. [39]

    Non-confusing generation of customized concepts in diffusion models

    Wang Lin, Jingyuan Chen, Jiaxin Shi, Yichen Zhu, Chen Liang, Junzhong Miao, Tao Jin, Zhou Zhao, Fei Wu, Shuicheng Yan, et al. Non-confusing generation of customized concepts in diffusion models. arXiv:2405.06914, 2024

    2024 arXiv

  32. [40]

    Multitwine: Multi-object compositing with text and layout control

    Zhe Lin, Zhifei Zhang, He Zhang, Andrew Gilbert, John Philip Collomosse, and Soo Ye Kim. Multitwine: Multi-object compositing with text and layout control. In CVPR, 2025

    2025

  33. [41]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Chunyuan Li, Jianwei Yang, Hang Su, Jun Zhu, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. arXiv:2303.05499, 2023. 12

    2023 arXiv

  34. [42]

    Step1x-edit: A practical framework for general image editing

    Shiyu Liu, Yucheng Han, Peng Xing, Fukun Yin, Rui Wang, Wei Cheng, Jiaqi Liao, Yingming Wang, Honghao Fu, Chunrui Han, et al. Step1x-edit: A practical framework for general image editing. arXiv:2504.17761, 2025

    2025 arXiv

  35. [43]

    Cones 2: Customizable image synthesis with multiple subjects

    Zhiheng Liu, Yifei Zhang, Yujun Shen, Kecheng Zheng, Kai Zhu, Ruili Feng, Yu Liu, Deli Zhao, Jingren Zhou, and Yang Cao. Cones 2: Customizable image synthesis with multiple subjects. In NeurIPS, pages 57500–57519, 2023

    2023

  36. [44]

    Customizable image synthesis with multiple subjects

    Zhiheng Liu, Yifei Zhang, Yujun Shen, Kecheng Zheng, Kai Zhu, Ruili Feng, Yu Liu, Deli Zhao, Jingren Zhou, and Yang Cao. Customizable image synthesis with multiple subjects. In NeurIPS, volume 36, pages 57500–57519, 2023

    2023

  37. [45]

    Tf-icon: Diffusion-based training-free cross-domain image composition

    Shilin Lu, Yanzhu Liu, and Adams Wai-Kin Kong. Tf-icon: Diffusion-based training-free cross-domain image composition. In ICCV, pages 2294–2305, 2023

    2023

  38. [46]

    Midjourney official website, 2025

    MidJourney. Midjourney official website, 2025

    2025

  39. [47]

    Knobgen: Controlling the sophistication of artwork in sketch-based diffusion models

    Pouyan Navard, Amin Karimi Monsefi, Mengxi Zhou, Wei-Lun Chao, Alper Yilmaz, and Rajiv Ramnath. Knobgen: Controlling the sophistication of artwork in sketch-based diffusion models. arXiv:2410.01595, 2024

    2024 arXiv

  40. [48]

    ChatGPT (model 4o)

    OpenAI. ChatGPT (model 4o). https://chat.openai.com/, 2025

    2025

  41. [49]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel HAZIZA, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. TMLR

  42. [50]

    Text2place: Affordance- aware text guided human placement

    Rishubh Parihar, Harsh Gupta, Sachidanand VS, and R Venkatesh Babu. Text2place: Affordance- aware text guided human placement. In ECCV, pages 57–77. Springer, 2024

    2024

  43. [51]

    Zero-shot image-to-image translation

    Gaurav Parmar, Krishna Kumar Singh, Richard Zhang, Yijun Li, Jingwan Lu, and Jun-Yan Zhu. Zero-shot image-to-image translation. In ACM SIGGRAPH, pages 1–11, 2023

    2023

  44. [52]

    λ-eclipse: Multi-concept personalized text-to-image diffusion models by leveraging clip latent space

    Maitreya Patel, Sangmin Jung, Chitta Baral, and Yezhou Yang. λ-eclipse: Multi-concept personalized text-to-image diffusion models by leveraging clip latent space. TMLR, 2024

    2024

  45. [53]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In ICCV, pages 4195–4205, 2023

    2023

  46. [54]

    Tale: Training-free cross-domain image compo- sition via adaptive latent manipulation and energy-guided optimization

    Kien T Pham, Jingye Chen, and Qifeng Chen. Tale: Training-free cross-domain image compo- sition via adaptive latent manipulation and energy-guided optimization. In ACM MM, pages 3160–3169, 2024

    2024

  47. [55]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. In ICML, pages 8748–8763, 2021

    2021

  48. [56]

    Sam 2: Segment anything in images and videos

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

    2024 arXiv

  49. [57]

    Relightful harmonization: Lighting-aware portrait background replacement

    Mengwei Ren, Wei Xiong, Jae Shin Yoon, Zhixin Shu, Jianming Zhang, HyunJoon Jung, Guido Gerig, and He Zhang. Relightful harmonization: Lighting-aware portrait background replacement. In CVPR, pages 6452–6462, 2024

    2024

  50. [58]

    Magic insert: Style-aware drag-and-drop

    Nataniel Ruiz, Yuanzhen Li, Neal Wadhwa, Yael Pritch, Michael Rubinstein, David E Jacobs, and Shlomi Fruchter. Magic insert: Style-aware drag-and-drop. arXiv:2407.02489, 2024

    2024 arXiv

  51. [59]

    Clic: Concept learning in context

    Mehdi Safaee, Aryan Mikaeili, Or Patashnik, Daniel Cohen-Or, and Ali Mahdavi-Amiri. Clic: Concept learning in context. In CVPR, pages 6924–6933, 2024

    2024

  52. [60]

    Dreamrelation: Bridging customization and relation generation

    Qingyu Shi, Lu Qi, Jianzong Wu, Jinbin Bai, Jingbo Wang, Yunhai Tong, and Xiangtai Li. Dreamrelation: Bridging customization and relation generation. In CVPR, 2025. 13

    2025

  53. [61]

    Zero123++: a single image to consistent multi-view diffusion base model

    Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang, Minghua Liu, Chao Xu, Xinyue Wei, Linghao Chen, Chong Zeng, and Hao Su. Zero123++: a single image to consistent multi-view diffusion base model. arXiv:2310.15110, 2023

    2023 arXiv

  54. [62]

    Insert anything: Image insertion via in-context editing in dit

    Wensong Song, Hong Jiang, Zongxing Yang, Ruijie Quan, and Yi Yang. Insert anything: Image insertion via in-context editing in dit. arXiv:2504.15009, 2025

    2025 arXiv

  55. [63]

    Objectstitch: Object compositing with diffusion model

    Yizhi Song, Zhifei Zhang, Zhe Lin, Scott Cohen, Brian Price, Jianming Zhang, Soo Ye Kim, and Daniel Aliaga. Objectstitch: Object compositing with diffusion model. In CVPR, pages 18310–18319, 2023

    2023

  56. [64]

    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 CVPR, pages 8048–8058, 2024

    2024

  57. [65]

    Motioncom: Automatic and motion-aware image composition with llm and video diffusion prior

    Weijing Tao, Xiaofeng Yang, Miaomiao Cui, and Guosheng Lin. Motioncom: Automatic and motion-aware image composition with llm and video diffusion prior. arXiv:2409.10090, 2024

    2024 arXiv

  58. [66]

    Thinking outside the bbox: Unconstrained generative object compositing

    Gemma Canet Tarrés, Zhe Lin, Zhifei Zhang, Jianming Zhang, Yizhi Song, Dan Ruta, Andrew Gilbert, John Collomosse, and Soo Ye Kim. Thinking outside the bbox: Unconstrained generative object compositing. arXiv:2409.04559, 2024

    2024 arXiv

  59. [67]

    Key-locked rank one editing for text-to-image personalization

    Yoad Tewel, Rinon Gal, Gal Chechik, and Yuval Atzmon. Key-locked rank one editing for text-to-image personalization. In ACM SIGGRAPH, pages 1–11, 2023

    2023

  60. [68]

    Mige: A unified framework for multimodal instruction-based image generation and editing

    Xueyun Tian, Wei Li, Bingbing Xu, Yige Yuan, Yuanzhuo Wang, and Huawei Shen. Mige: A unified framework for multimodal instruction-based image generation and editing. arXiv:2502.21291, 2025

    2025 arXiv

  61. [69]

    Sv3d: Novel multi-view synthesis and 3d generation from a single image using latent video diffusion

    Vikram V oleti, Chun-Han Yao, Mark Boss, Adam Letts, David Pankratz, Dmitry Tochilkin, Christian Laforte, Robin Rombach, and Varun Jampani. Sv3d: Novel multi-view synthesis and 3d generation from a single image using latent video diffusion. In ECCV, pages 439–457. Springer, 2024

    2024

  62. [70]

    Unicombine: Unified multi-conditional combination with diffusion transformer

    Haoxuan Wang, Jinlong Peng, Qingdong He, Hao Yang, Ying Jin, Jiafu Wu, Xiaobin Hu, Yanjie Pan, Zhenye Gan, Mingmin Chi, et al. Unicombine: Unified multi-conditional combination with diffusion transformer. arXiv:2503.09277, 2025

    2025 arXiv

  63. [71]

    Dreamactor-h1: High-fidelity human-product demonstration video generation via motion-designed diffusion transformers

    Lizhen Wang, Zhurong Xia, Tianshu Hu, Pengrui Wang, Pengfei Wang, Zerong Zheng, and Ming Zhou. Dreamactor-h1: High-fidelity human-product demonstration video generation via motion-designed diffusion transformers. arXiv preprint arXiv:2506.10568, 2025

    2025 arXiv

  64. [72]

    Ms-diffusion: Multi- subject zero-shot image personalization with layout guidance

    Xierui Wang, Siming Fu, Qihan Huang, Wanggui He, and Hao Jiang. Ms-diffusion: Multi- subject zero-shot image personalization with layout guidance. In ICLR, 2025

    2025

  65. [73]

    Magicface: Training-free universal-style human image customized synthesis

    Yibin Wang, Weizhong Zhang, and Cheng Jin. Magicface: Training-free universal-style human image customized synthesis. arXiv:2408.07433, 2024

    2024 arXiv

  66. [74]

    Primecomposer: Faster progres- sively combined diffusion for image composition with attention steering

    Yibin Wang, Weizhong Zhang, Jianwei Zheng, and Cheng Jin. Primecomposer: Faster progres- sively combined diffusion for image composition with attention steering. In ACM MM, pages 10824–10832, 2024

    2024

  67. [75]

    Genartist: Multimodal llm as an agent for unified image generation and editing

    Zhenyu Wang, Aoxue Li, Zhenguo Li, and Xihui Liu. Genartist: Multimodal llm as an agent for unified image generation and editing. In NeurIPS, volume 37, pages 128374–128395, 2024

    2024

  68. [76]

    Bovik, Hamid R

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

    2004

  69. [77]

    Dreamrelation: Relation-centric video customization

    Yujie Wei, Shiwei Zhang, Hangjie Yuan, Biao Gong, Longxiang Tang, Xiang Wang, Haonan Qiu, Hengjia Li, Shuai Tan, Yingya Zhang, et al. Dreamrelation: Relation-centric video customization. arXiv:2503.07602, 2025

    2025 arXiv

  70. [78]

    Objectdrop: Bootstrapping counterfactuals for photorealistic object removal and insertion

    Daniel Winter, Matan Cohen, Shlomi Fruchter, Yael Pritch, Alex Rav-Acha, and Yedid Hoshen. Objectdrop: Bootstrapping counterfactuals for photorealistic object removal and insertion. In ECCV, pages 112–129. Springer, 2024. 14

    2024

  71. [79]

    Objectmate: A recurrence prior for object insertion and subject-driven generation

    Daniel Winter, Asaf Shul, Matan Cohen, Dana Berman, Yael Pritch, Alex Rav-Acha, and Yedid Hoshen. Objectmate: A recurrence prior for object insertion and subject-driven generation. arXiv:2412.08645, 2024

    2024 arXiv

  72. [80]

    Flipconcept: Tuning-free multi-concept personalization for text-to-image generation

    Young Beom Woo and Sun Eung Kim. Flipconcept: Tuning-free multi-concept personalization for text-to-image generation. arXiv:2502.15203, 2025

    2025 arXiv

  73. [81]

    Grok 3: The age of reasoning agents

    xAI. Grok 3: The age of reasoning agents. https://x.ai/news/grok-3, 2025

    2025

  74. [82]

    Fastcom- poser: Tuning-free multi-subject image generation with localized attention

    Guangxuan Xiao, Tianwei Yin, William T Freeman, Frédo Durand, and Song Han. Fastcom- poser: Tuning-free multi-subject image generation with localized attention. IJCV, pages 1–20, 2024

    2024

  75. [83]

    Omnigen: Unified image generation

    Shitao Xiao, Yueze Wang, Junjie Zhou, Huaying Yuan, Xingrun Xing, Ruiran Yan, Chaofan Li, Shuting Wang, Tiejun Huang, and Zheng Liu. Omnigen: Unified image generation. In CVPR, 2025

    2025

  76. [84]

    Anchorcrafter: Animate cyberanchors saling your products via human-object interacting video generation

    Ziyi Xu, Ziyao Huang, Juan Cao, Yong Zhang, Xiaodong Cun, Qing Shuai, Yuchen Wang, Linchao Bao, Jintao Li, and Fan Tang. Anchorcrafter: Animate cyberanchors saling your products via human-object interacting video generation. arXiv:2411.17383, 2024

    2024 arXiv

  77. [85]

    Dccf: Deep comprehensible color filter learning framework for high-resolution image harmonization

    Ben Xue, Shenghui Ran, Quan Chen, Rongfei Jia, Binqiang Zhao, and Xing Tang. Dccf: Deep comprehensible color filter learning framework for high-resolution image harmonization. In ECCV, pages 300–316. Springer, 2022

    2022

  78. [86]

    Hoi-swap: Swapping objects in videos with hand-object interaction awareness

    Zihui Sherry Xue, Romy Luo, Changan Chen, and Kristen Grauman. Hoi-swap: Swapping objects in videos with hand-object interaction awareness. In NeurIPS, volume 37, pages 77132–77164, 2024

    2024

  79. [87]

    Paint by example: Exemplar-based image editing with diffusion models

    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 models. In CVPR, pages 18381–18391, 2023

    2023

  80. [88]

    Person in place: Generating associative skeleton-guidance maps for human-object interaction image editing

    ChangHee Yang, ChanHee Kang, Kyeongbo Kong, Hanni Oh, and Suk-Ju Kang. Person in place: Generating associative skeleton-guidance maps for human-object interaction image editing. In CVPR, pages 8164–8175, 2024

    2024

  81. [89]

    Effective whole-body pose estimation with two-stages distillation

    Zhendong Yang, Ailing Zeng, Chun Yuan, and Yu Li. Effective whole-body pose estimation with two-stages distillation. In ICCV, pages 4210–4220, 2023

    2023

  82. [90]

    Freegraftor: Training-free cross-image feature grafting for subject-driven text-to-image generation

    Zebin Yao, Lei Ren, Huixing Jiang, Chen Wei, Xiaojie Wang, Ruifan Li, and Fangxiang Feng. Freegraftor: Training-free cross-image feature grafting for subject-driven text-to-image generation. arXiv:2504.15958, 2025

    2025 arXiv

  83. [91]

    Affordance diffusion: Synthesizing hand-object interactions

    Yufei Ye, Xueting Li, Abhinav Gupta, Shalini De Mello, Stan Birchfield, Jiaming Song, Shubham Tulsiani, and Sifei Liu. Affordance diffusion: Synthesizing hand-object interactions. In CVPR, pages 22479–22489, 2023

    2023

  84. [92]

    Omnipaint: Mastering object- oriented editing via disentangled insertion-removal inpainting

    Yongsheng Yu, Ziyun Zeng, Haitian Zheng, and Jiebo Luo. Omnipaint: Mastering object- oriented editing via disentangled insertion-removal inpainting. arXiv:2503.08677, 2025

    2025 arXiv

  85. [93]

    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:2308.10040, 2023

    2023 arXiv

  86. [94]

    Efficient two-stage detection of human- object interactions with a novel unary-pairwise transformer

    Frederic Z Zhang, Dylan Campbell, and Stephen Gould. Efficient two-stage detection of human- object interactions with a novel unary-pairwise transformer. In CVPR, pages 20104–20112, 2022

    2022

  87. [95]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In ICCV, pages 3836–3847, 2023

    2023

  88. [96]

    Motioncrafter: One-shot motion customization of diffusion models

    Yuxin Zhang, Fan Tang, Nisha Huang, Haibin Huang, Chongyang Ma, Weiming Dong, and Changsheng Xu. Motioncrafter: One-shot motion customization of diffusion models. arXiv:2312.05288, 2023. 15

    2023 arXiv

  89. [97]

    A woman is holding an apple with one

    Yuxuan Zhang, Yiren Song, Jiaming Liu, Rui Wang, Jinpeng Yu, Hao Tang, Huaxia Li, Xu Tang, Yao Hu, Han Pan, et al. Ssr-encoder: Encoding selective subject representation for subject-driven generation. In CVPR, pages 8069–8078, 2024. 16 HOComp: Interaction-Aware Human-Object Co...

    2024

  90. [98]

    Please analyze and describe a suitable type of interaction between them and generate a simple prompt for this interaction

    Interaction Prompt Generation. The MLLM is queried with If and Ib using the instruction: “Please analyze and describe a suitable type of interaction between them and generate a simple prompt for this interaction. ”The model outputs a text prompt C describing the interaction type

  91. [99]

    Please describe the position of the foreground object and give bounding box coordinates so that it aligns with the specified interaction

    Object Box Prediction. Using If , Ib, and C, we query the MLLM with: “Please describe the position of the foreground object and give bounding box coordinates so that it aligns with the specified interaction. ”The model returns the object bounding box Bo

  92. [100]

    A woman is riding a horse

    Interaction Region Prediction. Given If , Ib, C, and Bo, we ask: “Based on the images and interaction prompt, and assuming the object is at Bo, identify the regions on the person that would be affected during the interaction and return their bounding box. ”The MLLM then predic...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.