Pith. sign in

REVIEW 3 major objections 6 minor 55 references

IPVTON: Image-based 3D Virtual Try-on with Image Prompt Adapter

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read IPVTON generates a rotatable 3D try-on from just a person photo and a garment photo.

desk verdict A plausible new combination of IP-Adapter with mask-guided prompts and a pseudo-silhouette loss for 3D try-on, but the headline outperformance claim rests on an 8-pair evaluation that omits the closest prior method. read the letter →

arxiv 2501.15616 v1 pith:LBIQLDNS submitted 2025-01-26 cs.CV

Xiaojing Zhong , Zhonghua Wu , Xiaofeng Yang , Guosheng Lin , Qingyao Wu This is my paper · ORCID
classification cs.CV
keywords image-based3Dvirtualtry-onscoredistillationsamplingimagepromptadaptermask-guidedattentionpseudosilhouettelossdiffusionmodelsSMPL-XDMTet
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

IPVTON claims to solve image-based 3D virtual try-on: from a single photo of a person and a single photo of a garment, it reconstructs a 3D human model that wears the garment and can be rotated to any viewpoint. The method combines a text-to-image diffusion model with an image prompt adapter, so the garment's visual features enter the diffusion prior through image-conditioned cross-attention. A mask-guided embedding confines those garment features to the try-on region, preserving the person's face and body elsewhere, and a pseudo silhouette loss drives the geometry to adopt the garment's outline. The paper reports that this scheme outperforms prior methods in both geometry faithfulness and texture faithfulness.

What carries the argument

The central machinery is the image prompt adapter, a cross-attention layer that injects image features into a text-to-image diffusion model, combined with a mask that multiplies the cross-attention output, forcing the garment's features to act only on the try-on region. Around this sits a two-stage optimization: geometry is driven by a normal-map SDS loss plus a pseudo silhouette loss that matches the rendered silhouette's mask and edges to a ControlNet-generated pseudo mask, and texture is driven by a color SDS loss plus a reconstruction loss on the non-target area. The 3D body itself is a DMTet hybrid representation initialized from the SMPL-X outer shell, so the optimization starts from a plausible human shape and only deforms within the shell's reachable space.

What would settle it

Render the final 3D model from the same front and back views used in Eqs. 6 and 7 and compare its silhouette and normal maps against the pseudo ground truths used during optimization; if the optimized model matches those pseudo labels but still visibly fails to reproduce the garment from the input photo, then the synthetic supervision itself is the bottleneck.

Watch

Extended reading notes

Core claim

On its own terms, IPVTON's central claim is that score distillation sampling with an image prompt adapter can turn two unpaired images into a dressed 3D human. The geometry is first optimized with a normal-map SDS loss whose image prompt is the estimated normal map of the garment, combined with a pseudo silhouette loss that compares the rendered mesh silhouette against a ControlNet-generated silhouette of the person wearing the garment. Mask-guided image prompt embeddings multiply the cross-attention output by a binary mask covering the try-on region, so garment features do not bleed into the rest of the body. Texture is then optimized with a similar masked SDS loss using the garment's color image, plus a reconstruction loss that pins down non-target areas to the source photo. The paper claims this produces 3D models that are faithful to both the garment's shape and texture while preserving the source identity, and shows quantitative and qualitative gains over TEXTure, TeCH, and IPDreamer.

Load-bearing premise

The load-bearing premise is that the ControlNet-generated pseudo silhouette and the ICON-derived pseudo normal maps are accurate enough to be treated as ground truth for the 3D geometry, so any error in those synthetic cues is inherited by the reconstructed garment shape.

Editorial extensions

If this is right

  • A data-efficient 3D try-on pipeline becomes possible: no paired human-garment training data and no per-sample fine-tuning are needed, only a human image, a garment image, and a text prompt.
  • Multi-angle assessment of a garment becomes possible for image-based input, overcoming the fixed-viewpoint limitation of 2D virtual try-on.
  • Mask-guided image prompt embeddings prevent the garment appearance from leaking into unrelated parts such as the face or hair.
  • The pseudo silhouette loss lets geometry reflect garment shapes even when text prompts are insufficient to describe them.
  • The combination of text and image prompts captures both high-level semantics and low-level appearance of the target garment.

Reading between the lines

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

  • Because the representation is built on SMPL-X, animating the resulting 3D model should be straightforward, but the paper does not demonstrate motion or pose transfer, so this is an untested extension.
  • For very loose garments the SMPL-X initialization limits the reachable shape space; one testable extension is to replace the outer-shell initialization with a more flexible shell or allow deformation outside the shell.
  • The mask-guided attention mechanism could be ported to other image-prompt 3D editing tasks where only a spatial region should be altered, such as face editing or localized object replacement.
  • If the pseudo silhouette is the quality bottleneck, a targeted experiment with garments whose silhouette clashes with the source pose would reveal how much the final result depends on ControlNet's reliability.
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

3 major / 6 minor

Summary. IPVTON proposes an image-based 3D virtual try-on method that, given a human image and an in-shop garment image, optimizes a hybrid DMTet/SMPL-X representation in two stages. Geometry is guided by normal-space SDS with IP-Adapter image prompts, mask-guided cross-attention, a pseudo silhouette loss, and a pseudo normal loss; texture is guided by image-prompt SDS plus a reconstruction loss on non-try-on regions. The paper reports qualitative results, CLIP-based faithfulness scores, a user study, and component ablations against TEXTure, TeCH, and IPDreamer.

Significance. If fully substantiated, IPVTON would be a useful data-efficient image-based 3D virtual try-on framework: it requires no paired training data, no per-sample fine-tuning, and it integrates garment shape and texture through a standard pre-trained IP-Adapter. The method design is coherent, the losses are conventional SDS and reconstruction objectives with no sign of circular evaluation, and the ablations in Figs. 5-6 support the qualitative roles of the mask-guided embeddings and the pseudo silhouette loss. The paper is also honest about known limitations (loose garments, complex patterns/logos). However, the central comparative claim of outperforming previous methods is not established by the current evidence, primarily because the closest prior method, DreamVTON, is cited but never compared against, and the quantitative evaluation is based on only 8 identity-garment pairs without variance or significance reporting.

major comments (3)
  1. [Experiments, Baselines and Table 1] The central claim that IPVTON outperforms previous image-based 3D virtual try-on methods is not supported by the baselines chosen. DreamVTON (Xie et al. 2024) is the closest prior method in this exact setting - it also uses SDS with a personalized diffusion model for image-based 3D try-on - and it is cited in Related Work; the paper even motivates IPVTON as avoiding DreamVTON's per-sample LoRA fine-tuning. Yet DreamVTON is absent from both Fig. 4 and Table 1. Without a direct qualitative and quantitative comparison to DreamVTON, the abstract's 'outperforms previous methods' claim cannot be assessed. Please add DreamVTON as a baseline, and report the training/optimization-time comparison that is needed to support the data-efficiency claim.
  2. [Quantitative Comparison, Table 1] The quantitative evaluation is too thin to support the claimed superiority. Only 8 identity-garment pairs are used; no standard deviations, confidence intervals, or significance tests are reported, so the CLIP-score gaps in Table 1 (e.g., Geo-Faith 31.77 vs. 31.41, Tex-Faith 33.60 vs. 32.34) may not be reliable. The CLIP scores are computed against text prompts, and geometry faithfulness is measured by prepending 'the normal map of' to a text prompt, which is a weak proxy for garment fidelity. The user study (15 volunteers, average ranks only) also lacks agreement or significance measures. Please expand the evaluation set, report per-pair results and variance, run significance tests, and add image-prompt-based faithfulness metrics that directly compare generated texture with the target garment image.
  3. [Eq. 6, Eq. 7, and Fig. 5 ablation] The geometry stage treats pseudo silhouettes from ControlNet/SAM and pseudo normal maps from ICON as ground truth, with lambda_PSL and lambda_norm both set to 10,000, but the paper's own ablation in Fig. 5 states that PSL alone can cause noisy seams and inaccurate shapes 'due to potential inaccuracies in the generated pseudo silhouette'. This is a load-bearing limitation for the geometry claim. The paper should quantify how often and under which garment types this failure occurs, analyze sensitivity to lambda_PSL and lambda_norm, and describe a mitigation strategy (e.g., confidence weighting of the pseudo supervision, or view selection beyond the front/back views). The Limitation section also concedes that extremely loose garments fail due to the SMPL-X shell, so the reachable geometry space should be delimited explicitly.
minor comments (6)
  1. [Eq. 6] Equation (6) contains a parenthesis mismatch: the first term is written as '\|H_M^{I'} - I_M)^2\|_2^2' with an extra closing parenthesis; please correct the notation.
  2. [Datasets vs. Quantitative Comparison] The paper states that 12 full-body human images are selected from DeepFashion, each paired with two garment templates, but the quantitative evaluation uses only 8 identity-garment pairs. Please clarify whether the 12 images refer to the qualitative set only and why only 8 are used for Table 1.
  3. [Method, masks] The mask notation is not fully consistent: Eq. (5) uses m for the try-on region mask, Eq. (11) uses m-hat for the complement mask, but the text does not define both masks explicitly at their first use. Please define m and m-hat precisely and state how they are derived from H_I and H'_I.
  4. [Implementation Details] The implementation details are under-specified: reporting only '100 iterations' is insufficient. Please include the optimizer, learning rate, number of rendered views per iteration, DMTet resolution, and total optimization time per sample, especially since the paper's efficiency argument is part of its contribution.
  5. [Datasets, references] The DeepFashion dataset is cited as '(Shen et al. 2021)', but the reference list entry with that author set is 'Deep Marching Tetrahedra'. Please use the correct DeepFashion citation.
  6. [Ablation study] The ablations in Figs. 5 and 6 are qualitative only. Adding quantitative ablation numbers (CLIP or other metrics, with/without each component) would make the component contributions more convincing.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the optimization objectives and evaluation metrics are independent, and self-citations are not load-bearing.

full rationale

I walked the paper's derivation chain and found no step in which a predicted quantity is defined in terms of the same quantity, no fitted parameter later reported as a prediction, and no load-bearing self-citation chain. The geometry and texture objectives (Eqs. 4, 6, 7, 9, 11) are standard SDS plus reconstruction, silhouette, and normal losses; the pseudo silhouette and pseudo normal targets come from external models (ControlNet, SAM, ICON, DPT), and although these are synthetic supervision, the paper's quantitative evaluation uses CLIP scores and user rankings on rendered views, not those same pseudo targets. IP-Adapter SDS transfers garment appearance and shape through a pre-trained external adapter, and the mask-guided embeddings are architectural choices rather than circular redefinitions. Self-citations (e.g., Zhong et al. 2023a, 2023b, 2025) appear only in related work or as general prior context and do not justify the central claim. The central comparative claim is weakened by the missing DreamVTON baseline and the small 8-pair evaluation scale, and the pseudo-silhouette supervision is acknowledged as potentially inaccurate in the paper's own ablation study and Limitation section, but these are evidence-quality concerns, not circularity. Therefore the derivation is self-contained with respect to circularity, and the score is 0.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard generative assumptions about SDS, IP-Adapter semantics, and the accuracy of synthetic geometry supervisors from ControlNet/SAM/ICON. The only hand-chosen numbers are the six loss weights, set without a validation search; the paper does not fit parameters to the test set, so circularity is low.

free parameters (6)
  • lambda_PSL = 10000
    Weight on the pseudo silhouette loss in Eq. 8, chosen by hand, not tuned against a validation metric.
  • lambda_norm = 10000
    Weight on the pseudo normal loss in Eq. 8, chosen by hand.
  • lambda_lap = 10000
    Weight on the Laplacian smoothing term in Eq. 8, chosen by hand.
  • lambda_norm_SDS = 1
    Weight on the normal SDS loss in Eq. 8, chosen by hand.
  • lambda_recon = 10000
    Weight on the texture reconstruction loss in Eq. 12, chosen by hand.
  • lambda_tex_SDS = 1
    Weight on the texture SDS loss in Eq. 12, chosen by hand.
assumptions (4)
  • domain assumption SMPL-X outer shell provides a valid initialization and shape prior for a clothed 3D human.
    The entire hybrid representation is initialized with SMPL-X; the Limitation section notes extremely loose garments fail, which shows the limits of this prior.
  • domain assumption Pseudo silhouette and pseudo normal maps generated by ControlNet, SAM, and ICON are accurate enough to serve as geometry ground truth.
    Eq. 6 and Eq. 7 minimize against these synthetic maps; the ablation in Fig. 5 shows PSL alone can create noisy seams when they are inaccurate.
  • domain assumption IP-Adapter image prompt embeddings capture garment semantics sufficiently for SDS-guided generation.
    IP-Adapter is the only channel for garment visual features; the Limitation section states complex patterns and logos are not captured.
  • domain assumption Score distillation sampling with a pretrained diffusion prior yields multi-view consistent 3D optimization.
    The method relies on SDS as established in prior 3D generation work; this is an assumption inherited from the literature, not proved in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of IPVTON: Image-based 3D Virtual Try-on with Image Prompt Adapter." pith.science (2026). https://pith.science/paper/LBIQLDNS

@misc{pith2026250115616,
  author       = {Pith},
  title        = {Pith review of: IPVTON: Image-based 3D Virtual Try-on with Image Prompt Adapter},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LBIQLDNS}},
  note         = {Machine review of arXiv:2501.15616}
}
read the original abstract

Given a pair of images depicting a person and a garment separately, image-based 3D virtual try-on methods aim to reconstruct a 3D human model that realistically portrays the person wearing the desired garment. In this paper, we present IPVTON, a novel image-based 3D virtual try-on framework. IPVTON employs score distillation sampling with image prompts to optimize a hybrid 3D human representation, integrating target garment features into diffusion priors through an image prompt adapter. To avoid interference with non-target areas, we leverage mask-guided image prompt embeddings to focus the image features on the try-on regions. Moreover, we impose geometric constraints on the 3D model with a pseudo silhouette generated by ControlNet, ensuring that the clothed 3D human model retains the shape of the source identity while accurately wearing the target garments. Extensive qualitative and quantitative experiments demonstrate that IPVTON outperforms previous methods in image-based 3D virtual try-on tasks, excelling in both geometry and texture.

Figures

Figures reproduced from arXiv: 2501.15616 by the authors.

Figure 1
Figure 1. Compared to 2D virtual try-on (Kim et al. 2024) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. 3D Try-on results. Given a human image, a garment image and a text prompt, IPVTON can generate realistic 3D human models with the desired garment shapes and textures while preserving the source identity. images. In this paper, we propose IPVTON, a data-efficient image-based 3D virtual try-on framework that integrates an image prompt adapter with customized diffusion models to optimize a hybrid 3D human model using S… view at source ↗
Figure 3
Figure 3. Overview of IPVTON. Given a human image HI , we first construct a DMTet-based 3d representation initialized with SMPL-X to model the human, with its geometry and texture generated through Ωg and Ωc, respectively. During geometry optimization, the rendered human normal map I n is encoded into the diffusion model ϵˆip and, along with yn and m, is used to compute L norm SDS . yn is the normal image prompt embedding enc… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons. Our IPVTON is able to generate realistic 3D try-on results with high-quality textures, viewable from multiple angles. Qualitative Comparison. As shown in [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Ablation study for geometry optimization. ‘mIPA‘ denotes mask-guided image prompt embeddings. Methods CLIP ↑ User ↑ Geo-Faith Tex-Faith Geometry Texture TEXTure 30.85 28.31 2.34 1.53 TeCH 31.41 32.34 2.55 2.6 IPDreamer 28.53 30.49 2.28 1.86 IPVTON 31.77 33.60 2.63 3.52…
Figure 6
Figure 6. Figure 6: Ablation study for texture optimization. ‘mIPA‘ denotes mask-guided image prompt embeddings. darker than the garment image. In the second column, when only the top is meant to be changed, the pants and hair are also affected if image prompt embeddings are used without …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 35 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Ando, R.; and Zhang, T. 2006. Learning on graph with Laplacian regularization. Advances in neural information processing systems, 19

    work page 2006

  4. [4]

    L.; Tiwari, G.; Theobalt, C.; and Pons-Moll, G

    Bhatnagar, B. L.; Tiwari, G.; Theobalt, C.; and Pons-Moll, G. 2019. Multi-garment net: Learning to dress 3d people from images. In Proceedings of the IEEE/CVF international conference on computer vision, 5420--5430

    work page 2019

  5. [5]

    Cao, Z.; Simon, T.; Wei, S.-E.; and Sheikh, Y. 2017. Realtime multi-person 2d pose estimation using part affinity fields. In Proceedings of the IEEE conference on computer vision and pattern recognition, 7291--7299

    work page 2017

  6. [6]

    Chen, R.; Chen, Y.; Jiao, N.; and Jia, K. 2023. Fantasia3d: Disentangling geometry and appearance for high-quality text-to-3d content creation. In Proceedings of the IEEE/CVF international conference on computer vision, 22246--22256

    work page 2023

  7. [7]

    Choi, S.; Park, S.; Lee, M.; and Choo, J. 2021. Viton-hd: High-resolution virtual try-on via misalignment-aware normalization. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 14131--14140

    work page 2021

  8. [8]

    Choi, Y.; Kwak, S.; Lee, K.; Choi, H.; and Shin, J. 2024. Improving diffusion models for virtual try-on. arXiv preprint arXiv:2403.05139

    arXiv 2024

Show all 55 references
  1. [9]

    Doi, A.; and Koide, A. 1991. An efficient method of triangulating equi-valued surfaces by using tetrahedral cells. IEICE TRANSACTIONS on Information and Systems, 74(1): 214--224

    1991

  2. [10]

    Feng, Y.; Choutas, V.; Bolkart, T.; Tzionas, D.; and Black, M. J. 2021. Collaborative regression of expressive bodies using moderation. In 2021 International Conference on 3D Vision (3DV), 792--804. IEEE

    2021

  3. [11]

    Ge, Y.; Song, Y.; Zhang, R.; Ge, C.; Liu, W.; and Luo, P. 2021. Parser-free virtual try-on via distilling appearance flows. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 8485--8493

    2021

  4. [12]

    Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; and Bengio, Y. 2020. Generative adversarial networks. Communications of the ACM, 63(11): 139--144

    2020

  5. [13]

    Ho, J.; Jain, A.; and Abbeel, P. 2020. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 6840--6851

    2020

  6. [14]

    Ho, J.; and Salimans, T. 2022. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598

    2022 arXiv

  7. [15]

    Hong, F.; Zhang, M.; Pan, L.; Cai, Z.; Yang, L.; and Liu, Z. 2022. Avatarclip: Zero-shot text-driven generation and animation of 3d avatars. arXiv preprint arXiv:2205.08535

    2022 arXiv

  8. [16]

    J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685

    2021 arXiv

  9. [17]

    Huang, X.; Shao, R.; Zhang, Q.; Zhang, H.; Feng, Y.; Liu, Y.; and Wang, Q. 2024 a . Humannorm: Learning normal diffusion model for high-quality and realistic 3d human generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 4568--4577

    2024

  10. [18]

    Huang, Y.; Wang, J.; Zeng, A.; Cao, H.; Qi, X.; Shi, Y.; Zha, Z.-J.; and Zhang, L. 2024 b . Dreamwaltz: Make a scene with complex 3d animatable avatars. Advances in Neural Information Processing Systems, 36

    2024

  11. [19]

    Huang, Y.; Yi, H.; Xiu, Y.; Liao, T.; Tang, J.; Cai, D.; and Thies, J. 2024 c . Tech: Text-guided reconstruction of lifelike clothed humans. In 2024 International Conference on 3D Vision (3DV), 1531--1542. IEEE

    2024

  12. [20]

    Kim, J.; Gu, G.; Park, M.; Park, S.; and Choo, J. 2024. Stableviton: Learning semantic correspondence with latent diffusion model for virtual try-on. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 8176--8185

    2024

  13. [21]

    Kumari, N.; Zhang, B.; Zhang, R.; Shechtman, E.; and Zhu, J.-Y. 2023. Multi-concept customization of text-to-image diffusion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1931--1941

    2023

  14. [22]

    Laine, S.; Hellsten, J.; Karras, T.; Seol, Y.; Lehtinen, J.; and Aila, T. 2020. Modular primitives for high-performance differentiable rendering. ACM Transactions on Graphics (ToG), 39(6): 1--14

    2020

  15. [23]

    Li, Y.; Chen, H.-y.; Larionov, E.; Sarafianos, N.; Matusik, W.; and Stuyck, T. 2024. Diffavatar: Simulation-ready garment optimization with differentiable simulation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 4368--4378

    2024

  16. [24]

    Liao, T.; Yi, H.; Xiu, Y.; Tang, J.; Huang, Y.; Thies, J.; and Black, M. J. 2024. Tada! text to animatable digital avatars. In 2024 International Conference on 3D Vision (3DV), 1508--1519. IEEE

    2024

  17. [25]

    Loper, M.; Mahmood, N.; Romero, J.; Pons-Moll, G.; and Black, M. J. 2023. SMPL: A skinned multi-person linear model. In Seminal Graphics Papers: Pushing the Boundaries, Volume 2, 851--866

    2023

  18. [26]

    Mir, A.; Alldieck, T.; and Pons-Moll, G. 2020. Learning to transfer texture from clothing images to 3d humans. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7023--7034

    2020

  19. [27]

    E.; Cai, Z.; Yang, L.; Tao, Q.; Wu, Z.; Zhang, T.; and Liu, Z

    Pang, H. E.; Cai, Z.; Yang, L.; Tao, Q.; Wu, Z.; Zhang, T.; and Liu, Z. 2024. Towards robust and expressive whole-body human pose and shape estimation. Advances in Neural Information Processing Systems, 36

    2024

  20. [28]

    Patel, C.; Liao, Z.; and Pons-Moll, G. 2020. Tailornet: Predicting clothing in 3d as a function of human pose, shape and garment style. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 7365--7375

    2020

  21. [29]

    A.; Tzionas, D.; and Black, M

    Pavlakos, G.; Choutas, V.; Ghorbani, N.; Bolkart, T.; Osman, A. A.; Tzionas, D.; and Black, M. J. 2019. Expressive body capture: 3d hands, face, and body from a single image. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 10975--10985

    2019

  22. [30]

    T.; and Mildenhall, B

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

    2022 arXiv

  23. [31]

    Qian, G.; Mai, J.; Hamdi, A.; Ren, J.; Siarohin, A.; Li, B.; Lee, H.-Y.; Skorokhodov, I.; Wonka, P.; Tulyakov, S.; et al. 2023. Magic123: One image to high-quality 3d object generation using both 2d and 3d diffusion priors. arXiv preprint arXiv:2306.17843

    2023 arXiv

  24. [32]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PMLR

    2021

  25. [33]

    Ran, L.; Cun, X.; Liu, J.-W.; Zhao, R.; Zijie, S.; Wang, X.; Keppo, J.; and Shou, M. Z. 2024. X-adapter: Adding universal compatibility of plugins for upgraded diffusion model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 8775--8784

    2024

  26. [34]

    Ranftl, R.; Bochkovskiy, A.; and Koltun, V. 2021. Vision transformers for dense prediction. In Proceedings of the IEEE/CVF international conference on computer vision, 12179--12188

    2021

  27. [35]

    Richardson, E.; Metzer, G.; Alaluf, Y.; Giryes, R.; and Cohen-Or, D. 2023. Texture: Text-guided texturing of 3d shapes. In ACM SIGGRAPH 2023 conference proceedings, 1--11

    2023

  28. [36]

    Ronneberger, O.; Fischer, P.; and Brox, T. 2015. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention--MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part II...

    2015

  29. [37]

    Ruiz, N.; Li, Y.; Jampani, V.; Pritch, Y.; Rubinstein, M.; and Aberman, K. 2023. 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, 22500--22510

    2023

  30. [38]

    Shen, T.; Gao, J.; Yin, K.; Liu, M.-Y.; and Fidler, S. 2021. Deep marching tetrahedra: a hybrid representation for high-resolution 3d shape synthesis. Advances in Neural Information Processing Systems, 34: 6087--6101

    2021

  31. [39]

    Shi, X.; Wu, Z.; Lin, G.; Cai, J.; and Joty, S. 2021. Remember what you have drawn: Semantic image manipulation with memory. arXiv preprint arXiv:2107.12579

    2021 arXiv

  32. [40]

    Sohl-Dickstein, J.; Weiss, E.; Maheswaranathan, N.; and Ganguli, S. 2015. Deep unsupervised learning using nonequilibrium thermodynamics. In International conference on machine learning, 2256--2265. PMLR

    2015

  33. [41]

    Song, Y.; and Ermon, S. 2019. Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems, 32

    2019

  34. [42]

    Wang, Z.; Lu, C.; Wang, Y.; Bao, F.; Li, C.; Su, H.; and Zhu, J. 2024. Prolificdreamer: High-fidelity and diverse text-to-3d generation with variational score distillation. Advances in Neural Information Processing Systems, 36

    2024

  35. [43]

    Wu, Z.; Lin, G.; Tao, Q.; and Cai, J. 2019. M2e-try on net: Fashion from model to everyone. In Proceedings of the 27th ACM international conference on multimedia, 293--301

    2019

  36. [44]

    Wu, Z.; Tao, Q.; Lin, G.; and Cai, J. 2020. Exploring bottom-up and top-down cues with attentive learning for webly supervised object detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 12936--12945

    2020

  37. [45]

    Xie, Z.; Dong, H.; Gao, Y.; Ma, Z.; and Liang, X. 2024. DreamVTON: Customizing 3D Virtual Try-on with Personalized Diffusion Models. arXiv preprint arXiv:2407.16511

    2024 arXiv

  38. [46]

    Xiu, Y.; Yang, J.; Tzionas, D.; and Black, M. J. 2022. Icon: Implicit clothed humans obtained from normals. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 13286--13296. IEEE

    2022

  39. [47]

    Ye, H.; Zhang, J.; Liu, S.; Han, X.; and Yang, W. 2023. Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models. arXiv preprint arXiv:2308.06721

    2023 arXiv

  40. [48]

    Zeng, B.; Li, S.; Feng, Y.; Li, H.; Gao, S.; Liu, J.; Li, H.; Tang, X.; Liu, J.; and Zhang, B. 2023. Ipdreamer: Appearance-controllable 3d object generation with image prompts. arXiv preprint arXiv:2310.05375

    2023 arXiv

  41. [49]

    Zhang, H.; Chen, B.; Yang, H.; Qu, L.; Wang, X.; Chen, L.; Long, C.; Zhu, F.; Du, D.; and Zheng, M. 2024. Avatarverse: High-quality & stable 3d avatar creation from text and pose. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 7124--7132

    2024

  42. [50]

    Zhao, F.; Xie, Z.; Kampffmeyer, M.; Dong, H.; Han, S.; Zheng, T.; Zhang, T.; and Liang, X. 2021. M3d-vton: A monocular-to-3d virtual try-on network. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 13239--13249

    2021

  43. [51]

    Zhong, X.; Huang, X.; Wu, Z.; Lin, G.; and Wu, Q. 2023 a . Sara: Controllable makeup transfer with spatial alignment and region-adaptive normalization. arXiv preprint arXiv:2311.16828

    2023 arXiv

  44. [52]

    Zhong, X.; Huang, X.; Yang, X.; Lin, G.; and Wu, Q. 2025. Deco: Decoupled human-centered diffusion video editing with motion consistency. In European Conference on Computer Vision, 352--370. Springer

    2025

  45. [53]

    Zhong, X.; Su, Y.; Wu, Z.; Lin, G.; and Wu, Q. 2023 b . DI-Net: Decomposed Implicit Garment Transfer Network for Digital Clothed 3D Human. arXiv preprint arXiv:2311.16818

    2023 arXiv

  46. [54]

    Zhong, X.; Wu, Z.; Tan, T.; Lin, G.; and Wu, Q. 2021. Mv-ton: Memory-based video virtual try-on network. In Proceedings of the 29th ACM International Conference on Multimedia, 908--916

    2021

  47. [55]

    Zhu, L.; Yang, D.; Zhu, T.; Reda, F.; Chan, W.; Saharia, C.; Norouzi, M.; and Kemelmacher-Shlizerman, I. 2023. Tryondiffusion: A tale of two unets. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 4606--4615

    2023

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.