Pith. sign in

REVIEW 4 major objections 6 minor 37 references

FashionComposer: Compositional Fashion Image Generation

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read FashionComposer claims that multiple garment images and an optional face can be composed into one realistic fashion image in a single diffusion pass, with subject-binding attention keeping each garment on the correct body region.

desk verdict One-pass multi-garment try-on with subject-binding attention is a genuine contribution, but the central compositional claim rests on a thin custom evaluation and an unvalidated token-isolation assumption. read the letter →

arxiv 2412.14168 v2 pith:UCFF3ZV3 submitted 2024-12-18 cs.CV

Sihui Ji , Yiyang Wang , Xi Chen , Xiaogang Xu , Hao Luo , Hengshuang Zhao This is my paper · ORCID
classification cs.CV
keywords compositionalfashionimagegenerationsubject-bindingattentionreferenceUNetvirtualtry-onmulti-garmentcompositionhumanalbumdiffusionmodelsmulti-modalconditioning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that a single diffusion model can compose any number of garment images and an optional face into one fashion photo in a single pass, with each garment keeping its patterns and textures and appearing on the correct body region. The key to this is subject-binding attention, which links the visual features of each garment in the reference image to the text phrase that describes that garment. If the claim holds, virtual try-on moves from single-item fitting to full-outfit composition, and the approach may extend to other kinds of multi-object image generation.

What carries the argument

Subject-binding attention, defined in Eq. (3), is the central mechanism: for each asset region in the reference UNet's self-attention feature map, the downsampled tokens $k_j$ are replaced by $k'_j = \text{MLP}_l(P_i) + k_j$, where $P_i$ is the text embedding of the phrase that names that asset. This adds semantic information to the appearance tokens before they are concatenated into the denoising UNet's self-attention, allowing the model to bind each garment's appearance to the correct textual description and to the correct body pixels. The reference UNet (following the design of cited reference-UNet work) supplies the high-fidelity appearance features, and the asset-library arrangement keeps the number of reference UNets at one regardless of how many garments are present.

What would settle it

Take two garments of nearly the same color and texture, place them adjacent in the asset image, and check whether the generated image consistently places each garment on the correct body region across many random seeds; if the garments swap or blend appearance (e.g., the shirt picks up the skirt's pattern), the binding has failed and the central claim is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that multi-reference fashion images can be generated without fine-tuning by arranging all references in one 'asset library' image, extracting their features with a reference UNet, and using subject-binding attention to shift each asset's key/value tokens toward the text embedding of its matching phrase. This lets the model know which visual region corresponds to 'a red shirt' versus 'a blue skirt,' so each reference is transferred to the right body area while the reference UNet preserves fine detail. The paper further reports that the framework handles standard virtual try-on, multi-garment try-on, and human album generation with identity consistency, and that its ablations show binding on all UNet blocks gives the best balance of quality and fidelity. The authors also acknowledge that the model's generation capacity is limited by the scale and bias of the training data in terms of race, gender, and body figure.

Load-bearing premise

The method assumes that downsampling each asset's region in the reference feature map isolates only the tokens of that asset, so that adding the right text embedding attaches the right appearance to the right body part.

Editorial extensions

If this is right

  • Virtual try-on becomes a single-pass operation for full outfits, including garments that existing methods cannot condition on, such as lower-body clothes and shoes.
  • The cost of adding more reference garments does not grow with the number of garments, since all references share one asset image and one reference UNet.
  • Users can control pose, body figure, and viewpoint through a parametric model and a text prompt simultaneously, opening up try-on beyond fixed photos.
  • Human album generation with consistent identity follows from the same framework using cross-frame attention, correspondence-aware attention, and latent code alignment.

Reading between the lines

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

  • If subject-binding attention generalizes beyond fashion, the same asset-library plus text-binding recipe could enable multi-object customization in other domains (e.g., interior design, product images) as long as the objects can be placed in one reference collage. This is my inference, not stated by the paper.
  • The token-assignment step assumes downsampling cleanly separates assets; when assets touch or overlap in the asset image, the binding may degrade. A stress test on overlapping references would clarify the practical limit of the method.
  • The method could likely be combined with other conditioning signals such as depth or edge maps, since the asset-library and attention-binding modules are decoupled from the specific pose, but this is not explored 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 / 6 minor

Summary. FashionComposer is a diffusion-based framework for compositional fashion image generation. It accepts a text prompt, a parametric human model (SMPL/densepose), and multiple visual references (garment images, optional face) arranged in a single 'asset library' image. A reference UNet extracts appearance features, and a proposed 'subject-binding attention' (Eq. 3) adds text-phrase embeddings to the key/value tokens of each asset's spatial region, so that each garment's appearance is bound to the corresponding noun phrase and injected into the correct body region during denoising. The paper reports strong standard virtual try-on results on VITON-HD (Table 2), qualitative and quantitative comparisons for multi-garment customization (Table 1, Figures 3-4), ablations of the reference UNet and subject-binding attention (Tables 3-4, Figures 6-7), and a human album generation application with correspondence-aware attention and latent code alignment (Figure 8). The claimed contributions are a unified multi-modal conditioning framework, a scalable subject-binding mechanism for multiple references in one pass, and supporting techniques for consistent identity generation.

Significance. If the central claim holds, FashionComposer would be a meaningful advance over single-garment virtual try-on and multi-subject customization methods: it supports multiple garments and an optional face in a single diffusion pass without per-sample fine-tuning, while retaining detailed garment fidelity. The paper's strengths include a coherent framework, a large automatically constructed training set (165k samples), a strong benchmark result on VITON-HD where the method outperforms several strong baselines on FID/KID and SSIM, and a clean ablation showing that the reference UNet and subject-binding attention each contribute to fidelity and identity separation. The paper is weakened by the evaluation of its core compositional claim: the evidence for multi-garment compositionality is a 100-prompt comparison with global image-level metrics, a small user study for the binding ablation, and a qualitative section whose baseline inputs are adjusted on a best-effort basis. The central mechanism, subject-binding attention, is not stress-tested for token isolation, which is the main correctness risk.

major comments (4)
  1. [§3.4, Eq. (3)]
  2. [§4.2, Table 1]
  3. [§4.2, Figures 3-4]
  4. [§4.4, Table 4]
minor comments (6)
  1. [Abstract / §1]
  2. [§2 / Related Work]
  3. [§3.2, Figure 2]
  4. [§4.1]
  5. [§5, Conclusion]
  6. [Throughout]

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FashionComposer's central claims are empirical and benchmarked against external methods and datasets, with subject-binding attention defined as a concrete mechanism rather than derived from the output.

full rationale

FashionComposer is an empirical system paper rather than a derivation with fitted constants used to produce a prediction. The central technical construction, subject-binding attention (Section 3.4, Eq. 3), is defined as k'_j = MLP_l(P_i) + k_j: selected key tokens from an asset's bounding-box region are augmented with text-dependent MLP outputs. This is a mechanism design, not a tautology, and the paper's multi-garment fidelity and compositionality claims are supported by comparisons with external baselines (Emu2, Collage Diffusion, AnyDoor, Paint by Example) in Tables 1 and 3, standard benchmarks (VITON-HD, DressCode) in Table 2, and external metrics (CLIP-I, DINO, CLIP-T). The only self-referential element is the adoption of the reference-UNet design based on prior work [3, 5, 13, 36], some by overlapping authors, but this is a design choice and not a load-bearing inference: the reported numbers are measured on external datasets, not computed from those cited papers, and no uniqueness theorem or renamed fitted parameter is invoked. Concerns such as token mixing when asset boxes are close together or at different scales are correctness risks under Eq. 3, not circularity, because a failure would make the empirical claims false rather than true by construction. No self-definitional step, fitted-input-as-prediction step, or self-citation chain forcing the result was found.

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

This is an empirical ML system, not a derivation, so there are no numeric constants fitted to force agreement with a target result. The central claim instead rests on pre-trained components, automatically generated training annotations, and the assumption that downsampling an asset's region isolates that asset's tokens. These assumptions are plausible but not independently verified.

assumptions (4)
  • domain assumption Stable Diffusion v1.5 provides a usable generative prior for fashion imagery.
    The pipeline is built as fine-tuning of SD v1.5; if the backbone cannot render garments, the composition claim fails. Invoked in Sections 3.1 and 3.2.
  • domain assumption Mask2Former and Qwen models produce accurate garment masks, captions, and phrase-component links for 165k training samples.
    Training data construction in Section 3.3 depends on these off-the-shelf models; errors propagate to subject-binding text alignment.
  • domain assumption Densepose maps from SMPL are a sufficient pose and figure control signal when concatenated to the noise latent.
    Section 3.3 chooses this simple conditioning to avoid complex pose encoders; if pose control is insufficient, the generated body may not follow the SMPL input.
  • ad hoc to paper Downsampling an asset's region in the reference UNet feature map isolates that asset's tokens, and the text prompt contains a phrase for every asset.
    This is the core mechanism of subject-binding attention (Eq. 3), and it is not validated in isolation, only through the full model's outputs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FashionComposer: Compositional Fashion Image Generation." pith.science (2026). https://pith.science/paper/UCFF3ZV3

@misc{pith2026241214168,
  author       = {Pith},
  title        = {Pith review of: FashionComposer: Compositional Fashion Image Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UCFF3ZV3}},
  note         = {Machine review of arXiv:2412.14168}
}
read the original abstract

We present FashionComposer for compositional fashion image generation. Unlike previous methods, FashionComposer is highly flexible. It takes multi-modal input (i.e., text prompt, parametric human model, garment image, and face image) and supports personalizing the appearance, pose, and figure of the human and assigning multiple garments in one pass. To achieve this, we first develop a universal framework capable of handling diverse input modalities. We construct scaled training data to enhance the model's robust compositional capabilities. To accommodate multiple reference images (garments and faces) seamlessly, we organize these references in a single image as an "asset library" and employ a reference UNet to extract appearance features. To inject the appearance features into the correct pixels in the generated result, we propose subject-binding attention. It binds the appearance features from different "assets" with the corresponding text features. In this way, the model could understand each asset according to their semantics, supporting arbitrary numbers and types of reference images. As a comprehensive solution, FashionComposer also supports many other applications like human album generation, diverse virtual try-on tasks, etc.

Figures

Figures reproduced from arXiv: 2412.14168 by the authors.

Figure 1
Figure 1. Demonstration for the applications of FashionComposer. FashionComposer takes different kinds of conditions (e.g., garment image, face image, parametric human model) equally as “assets” to composite diverse and realistic fashion images. Thus supporting various fashion-related applications like controllable model image generation, virtual try-on, human album generation, etc. Abstract We present FashionComposer for com… view at source ↗
Figure 2
Figure 2. Overall pipeline of FashionComposer. FashionComposer takes garments composition and optional face, text prompt, and a densepose map projected from SMPL as inputs. The text prompt is encoded and fused with UNets through cross-attention and subject￾binding attention, while the garment features are extracted and injected for denoising through Feature Injection Attention. 3. Method 3.1. Preliminaries Stable Diffusion. W… view at source ↗
Figure 3
Figure 3. Qualitative comparison with multi-reference cus￾tomization methods, including Emu2 [27], Collage Diffu￾sion [25], Paint by Example [34] and AnyDoor [6]. the MLP used in text augmentation, and the convolution-in layer of the denoising UNet. Evaluation metrics. We evaluate the similarity between the synthesized and the reference garments by calculating the CLIP-Score (CLIP-I) and DINO-Score following Dream￾Booth [24].… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison with garment-centric fashion image synthesis methods, including StableGarment [30], IMAGDressing-v1 [26], and Magic Clothing [4], where ours better preserves the identity of the target objects. Note that all approaches do not finetune the model o…
Figure 5
Figure 5. Figure 5: Diverse virtual try-on results of FashionComposer for upper, lower, and outfit try-on tasks [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Qualitative comparison for the reference encoder. Reference UNet better preserves the fine details of the garments. cus more on ensuring structural and perceptual similarity but lack realism and detailed naturalness than our diffusion￾based model, which convincingly ve…
Figure 7
Figure 7. Figure 7: Qualitative ablation study on subject-binding atten￾tion. Bind(1) means only modifying the self-attention modules of UNet blocks with the smallest resolution. Conv-in refers to inject￾ing the mask map through the Convolution-in layer of the refer￾ence UNet. We highligh…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 18 canonical work pages

  1. [1]

    Qwen technical report

    Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, et al. Qwen technical report. arXiv:2309.16609, 2023. 4

    arXiv 2023

  2. [2]

    Qwen-vl: A versatile vision-language model for understanding, localization, text reading, and beyond

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. Qwen-vl: A versatile vision-language model for understanding, localization, text reading, and beyond. arXiv:2308.12966, 2023. 2, 4

    arXiv 2023

  3. [3]

    Wear-any-way: Manipulable virtual try-on via sparse correspondence alignment

    Mengting Chen, Xi Chen, Zhonghua Zhai, Chen Ju, Xuewen Hong, Jinsong Lan, and Shuai Xiao. Wear-any-way: Manipulable virtual try-on via sparse correspondence alignment. arXiv:2403.12965, 2024. 4

    arXiv 2024

  4. [4]

    Magic clothing: Controllable garment-driven image synthesis

    Weifeng Chen, Tao Gu, Yuhao Xu, and Chengcai Chen. Magic clothing: Controllable garment-driven image synthesis. arXiv:2404.09512, 2024. 6

    arXiv 2024

  5. [5]

    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. arXiv:2406.07547, 2024. 4

    arXiv 2024

  6. [6]

    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, 2024. 5

    work page 2024

  7. [7]

    Schwing, Alexander Kirillov, and Rohit Girdhar

    Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, and Rohit Girdhar. Masked-attention mask transformer for universal image segmentation. In CVPR, 2021. 2, 4

    work page 2021

  8. [8]

    Improving diffusion models for virtual try-on

    Yisol Choi, Sangkyung Kwak, Kyungmin Lee, Hyungwon Choi, and Jinwoo Shin. Improving diffusion models for virtual try-on. arXiv:2403.05139, 2024. 2

    arXiv 2024

Show all 37 references
  1. [9]

    An image is worth one word: Personalizing text-to-image generation using textual inversion

    Rinon Gal, Yuval Alaluf, Yuval Atzmon, Or Patashnik, Amit H Bermano, Gal Chechik, and Daniel Cohen-Or. An image is worth one word: Personalizing text-to-image generation using textual inversion. In ICLR, 2023. 2, 5

    2023

  2. [10]

    Taming the power of diffusion models for high-quality virtual try-on with appearance flow

    Junhong Gou, Siyu Sun, Jianfu Zhang, Jianlou Si, Chen Qian, and Liqing Zhang. Taming the power of diffusion models for high-quality virtual try-on with appearance flow. In ACMMM, 2023. 7

    2023

  3. [11]

    Densepose: Dense human pose estimation in the wild

    Rıza Alp G ¨uler, Natalia Neverova, and Iasonas Kokkinos. Densepose: Dense human pose estimation in the wild. In CVPR, 2018. 2

    2018

  4. [12]

    Viton: An image-based virtual try-on network

    Xintong Han, Zuxuan Wu, Zhe Wu, Ruichi Yu, and Larry S Davis. Viton: An image-based virtual try-on network. In CVPR, 2018. 2, 4, 6, 7

    2018

  5. [13]

    Animate anyone: Consistent and controllable image-to-video synthesis for character animation

    Li Hu, Xin Gao, Peng Zhang, Ke Sun, Bang Zhang, and Liefeng Bo. Animate anyone: Consistent and controllable image-to-video synthesis for character animation. arXiv:2311.17117, 2023. 2, 4

    2023 arXiv

  6. [14]

    Text2video-zero: Text-to-image diffusion models are zero-shot video generators

    Levon Khachatryan, Andranik Movsisyan, Vahram Tadevosyan, Roberto Henschel, Zhangyang Wang, Shant Navasardyan, and Humphrey Shi. Text2video-zero: Text-to-image diffusion models are zero-shot video generators. arXiv:2303.13439, 2023. 5

    2023 arXiv

  7. [15]

    Stableviton: Learning semantic correspondence with latent diffusion model for virtual try-on

    Jeongho Kim, Guojung Gu, Minho Park, Sunghyun Park, and Jaegul Choo. Stableviton: Learning semantic correspondence with latent diffusion model for virtual try-on. In CVPR, 2024. 2, 7

    2024

  8. [16]

    Deepfashion: Powering robust clothes recognition and retrieval with rich annotations

    Ziwei Liu, Ping Luo, Shi Qiu, Xiaogang Wang, and Xiaoou Tang. Deepfashion: Powering robust clothes recognition and retrieval with rich annotations. In CVPR, 2016. 4

    2016

  9. [17]

    Cones: Concept neurons in diffusion models for customized generation

    Zhiheng Liu, Ruili Feng, Kai Zhu, Yifei Zhang, Kecheng Zheng, Yu Liu, Deli Zhao, Jingren Zhou, and Yang Cao. Cones: Concept neurons in diffusion models for customized generation. arXiv:2303.05125, 2023. 2

    2023 arXiv

  10. [18]

    Smpl: A skinned multi-person linear model

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons-Moll, and Michael J Black. Smpl: A skinned multi-person linear model. In Seminal Graphics Papers: Pushing the Boundaries. 2023. 2

    2023

  11. [19]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv:1711.05101, 2017. 5

    2017 arXiv

  12. [20]

    Dress Code: High-Resolution Multi-Category Virtual Try-On

    Davide Morelli, Matteo Fincato, Marcella Cornia, Federico Landi, Fabio Cesari, and Rita Cucchiara. Dress Code: High-Resolution Multi-Category Virtual Try-On. In ECCV,

  13. [21]

    Ladi-vton: Latent diffusion textual-inversion enhanced virtual try-on

    Davide Morelli, Alberto Baldrati, Giuseppe Cartella, Marcella Cornia, Marco Bertini, and Rita Cucchiara. Ladi-vton: Latent diffusion textual-inversion enhanced virtual try-on. In ACMMM, 2023. 2, 7

    2023

  14. [22]

    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, 2021. 2

    2021

  15. [23]

    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 CVPR, 2022. 3

    2022

  16. [24]

    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 CVPR, 2023. 2, 5

    2023

  17. [25]

    Collage diffusion

    Vishnu Sarukkai, Linden Li, Arden Ma, Christopher R ´e, and Kayvon Fatahalian. Collage diffusion. In WACV, 2024. 2, 5

    2024

  18. [26]

    Imagdressing-v1: Customizable virtual dressing

    Fei Shen, Xin Jiang, Xin He, Hu Ye, Cong Wang, Xiaoyu Du, Zechao Li, and Jinhui Tang. Imagdressing-v1: Customizable virtual dressing. arXiv:2407.12705, 2024. 6

    2024 arXiv

  19. [27]

    Generative multimodal models are in-context learners

    Quan Sun, Yufeng Cui, Xiaosong Zhang, Fan Zhang, Qiying Yu, Zhengxiong Luo, Yueze Wang, Yongming Rao, Jingjing Liu, Tiejun Huang, et al. Generative multimodal models are in-context learners. arXiv:2312.13286, 2023. 2, 5

    2023 arXiv

  20. [28]

    Toward characteristic-preserving image-based virtual try-on network

    Bochao Wang, Huabin Zheng, Xiaodan Liang, Yimin Chen, Liang Lin, and Meng Yang. Toward characteristic-preserving image-based virtual try-on network. In ECCV, 2018. 2

    2018

  21. [29]

    Mv-vton: Multi-view virtual try-on with diffusion models

    Haoyu Wang, Zhilu Zhang, Donglin Di, Shiliang Zhang, and Wangmeng Zuo. Mv-vton: Multi-view virtual try-on with diffusion models. arXiv:2404.17364, 2024. 2, 7

    2024 arXiv

  22. [30]

    Stablegarment: Garment-centric generation via stable diffusion

    Rui Wang, Hailong Guo, Jiaming Liu, and Huaxia Li. Stablegarment: Garment-centric generation via stable diffusion. arXiv:2403.10783, 2024. 6, 7 9

    2024 arXiv

  23. [31]

    Fastcomposer: Tuning-free multi-subject image generation with localized attention

    Guangxuan Xiao, Tianwei Yin, William T Freeman, Fr ´edo Durand, and Song Han. Fastcomposer: Tuning-free multi-subject image generation with localized attention. arXiv:2305.10431, 2023. 2

    2023 arXiv

  24. [32]

    Gp-vton: Towards general purpose virtual try-on via collaborative local-flow global-parsing learning

    Zhenyu Xie, Zaiyu Huang, Xin Dong, Fuwei Zhao, Haoye Dong, Xijin Zhang, Feida Zhu, and Xiaodan Liang. Gp-vton: Towards general purpose virtual try-on via collaborative local-flow global-parsing learning. In CVPR,

  25. [33]

    Ootdiffusion: Outfitting fusion based latent diffusion for controllable virtual try-on

    Yuhao Xu, Tao Gu, Weifeng Chen, and Chengcai Chen. Ootdiffusion: Outfitting fusion based latent diffusion for controllable virtual try-on. arXiv:2403.01779, 2024. 2, 7

    2024 arXiv

  26. [34]

    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, 2023. 5

    2023

  27. [35]

    Deep learning technique for human parsing: A survey and outlook

    Lu Yang, Wenhe Jia, Shan Li, and Qing Song. Deep learning technique for human parsing: A survey and outlook. arXiv:2301.00394, 2023. 4

    2023 arXiv

  28. [36]

    Flashface: Human image personalization with high-fidelity identity preservation

    Shilong Zhang, Lianghua Huang, Xi Chen, Yifei Zhang, Zhi-Fan Wu, Yutong Feng, Wei Wang, Yujun Shen, Yu Liu, and Ping Luo. Flashface: Human image personalization with high-fidelity identity preservation. arXiv:2403.17008,

  29. [37]

    Tryondiffusion: A tale of two unets

    Luyang Zhu, Dawei Yang, Tyler Zhu, Fitsum Reda, William Chan, Chitwan Saharia, Mohammad Norouzi, and Ira Kemelmacher-Shlizerman. Tryondiffusion: A tale of two unets. In CVPR, 2023. 2 10

    2023

Pith tools

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