Pith. sign in

REVIEW 4 major objections 6 minor 52 references

WearWow claims the first end-to-end, mask-free framework for native 2K multi-garment virtual try-on, enabled by adaptive token packing and a dual-reward preference alignment that rescues high-frequency fabric texture.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 11:17 UTC pith:ONXDQYDW

load-bearing objection Plausible system, competitive single-garment numbers, but the multi-garment SOTA is built on a self-created dataset whose validation is deferred to a missing supplementary — that claim is unverified as it stands. the 4 major comments →

arxiv 2607.19923 v1 pith:ONXDQYDW submitted 2026-07-22 cs.CV

WearWow: Native 2K Multi-Garment Virtual Try-On via Adaptive Token Packing and Preference Alignment

Xujie Zhang , Runyan Du , Song Chang , Jiang Li , Dongliang Shao , Liping Wu , Wei Luo , Xiaochao Qu
show 2 more authors
Luoqi Liu Xiaodan Liang
This is my paper
classification cs.CV
keywords virtual try-on2K resolutionmulti-garment synthesismask-free generationtoken packingpreference alignmentdiffusion modelstexture fidelity
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

WearWow sets out to make native 2K multi-garment virtual try-on practical and realistic in one end-to-end, mask-free system. The paper argues that two obstacles block the field: the quadratic memory blow-up from passing many high-resolution garment images into a diffusion transformer, and the MSE denoising objective's spectral bias, which smooths away micro-textures like wool and denim at 2K. Its answer is Adaptive 2D Token Packing, which reorganizes garment images into a compact 2D canvas and prunes background tokens, plus a Multi-dimensional Try-on Reward, which aligns generation away from over-smoothed outputs using both a distribution-anchoring loss and semantic text guidance. If correct, WearWow would be the first open vertical system to make native 2K multi-garment try-on feasible with texture-faithful, mask-free generation, and the release of the WearWow-2K dataset would provide the first benchmark for it.

Core claim

WearWow is a generative diffusion framework claiming state-of-the-art results for native 2K (2048×1536) multi-garment virtual try-on without user-provided masks. It identifies two scaling barriers: self-attention cost grows as O((N_base + K·N)²) when K reference garments are concatenated, making 2K multi-garment training infeasible on current hardware; and the MSE-based diffusion objective acts as a low-pass filter at high resolution, producing plastic-like fabrics. The framework answers with ATP, which packs heterogeneous garment images onto a single 2D latent canvas by 2D bin-packing and then prunes background void tokens, cutting the effective sequence length while preserving 2D positiona

What carries the argument

Adaptive 2D Token Packing (ATP) — an algorithmic step that treats each reference garment as a 2D item, packs K items onto a unified canvas by geometric scaling and 2D bin-packing, then uses valid-region masks to prune empty background tokens after patchification; it preserves the original 2D spatial priors via 2D positional encodings and bounds attention cost to O((N_base+L−m)²). Multi-dimensional Try-on Reward (MTR) — a post-training alignment objective combining Cloth Distribution Reward (CDR), an implicit reward defined by the difference between trainable and frozen-reference velocity prediction errors on chosen vs rejected noisy samples, and Semantic Guidance Reward (SGR), a margin-based

Load-bearing premise

The load-bearing premise is that WearWow-2K's 'clothing-agnostic' target images, produced by an automated try-off pipeline and asserted to be physically correct and flawless (details deferred to the supplementary), are faithful stand-ins for real user photos; if they contain artifacts or distributional bias, both mask-free generation and the reported 2K results will not transfer to real-world inputs.

What would settle it

Run the released checkpoint on real (non-synthetic) paired photos where the same person is photographed both with and without multi-garment outfits, then compare structural and textural metrics; alternatively, measure the distribution gap between synthesized agnostic images and real agnostic images — a large gap would invalidate the mask-free claim.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • End-to-end training at native 2K with multiple garment conditions becomes computationally tractable, removing the need for cascaded super-resolution and its structural hallucinations.
  • Micro-level fabric texture (wool, denim, fleece) can be recovered at high resolution by combining distribution-anchored and semantic-guided preference alignment, counteracting MSE spectral bias.
  • Mask-free inference works: the model learns occlusion and depth ordering directly from synthesized agnostic training targets, so users do not need to provide segmentation masks.
  • The WearWow-2K dataset (≈100k native 2K triplets, 1–6 items, including footwear and accessories) becomes a benchmark for ultra-HD multi-garment try-on evaluation.
  • On both single-garment and multi-garment tests, WearWow reports better FID, KID, LPIPS, SSIM and human preference scores than existing open and commercial baselines.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If MTR's dual-reward design is the real cause of texture recovery, the same recipe (implicit distribution reward plus semantic margin loss) could transfer to other high-resolution generative tasks where MSE flattening is a problem, such as material synthesis or video super-resolution.
  • ATP's prune-and-pack idea generalizes to any conditional input with large uninformative regions, not just flat-lay garments; the memory savings should scale with foreground sparsity rather than image count alone.
  • The central risk is domain transfer: because both training and the 2K test set come from the same synthetic try-off pipeline, a real-user study with true paired before/after photos is the natural next validation. A large distribution gap between synthesized agnostic images and real photos would weaken the mask-free claim.
  • Independent reproduction should report actual peak VRAM as a function of K at 2K; the paper's complexity bound is plausible but the hardware numbers are not shown, and real-world multi-garment prompts often exceed the 1–6 item range of the dataset.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes WearWow, an end-to-end, mask-free virtual try-on framework operating natively at 2K resolution for multiple garments. To handle the memory cost of many high-resolution reference conditions, it introduces Adaptive 2D Token Packing (ATP), which packs garment images onto a shared 2D canvas and prunes background tokens. To counteract the over-smoothing of high-frequency fabric textures, it proposes a Multi-dimensional Try-on Reward (MTR) that combines a Cloth Distribution Reward (CDR), based on DPO-style preference pairs, with a Semantic Guidance Reward (SGR) that uses contrastive text anchors. The authors also introduce WearWow-2K, a 100k/2k training/test dataset of native 2K multi-garment triplets with synthesized clothing-agnostic person images. The paper reports state-of-the-art results on WearWow-2K for multi-garment synthesis against commercial APIs and competitive single-garment results on VITON-HD and DressCode.

Significance. If the central claim is validated, WearWow would be a meaningful advance: it directly addresses two real bottlenecks in high-resolution multi-garment try-on — the quadratic attention memory blow-up and the spectral bias of diffusion objectives — and its ATP/MTR formulations are well-motivated and clearly described. The single-garment external benchmarks (VITON-HD, DressCode) give partial independent grounding and show competitive performance. The paper is also careful in ablating its two main components. However, the multi-garment SOTA claim rests entirely on WearWow-2K, whose synthesized agnostic targets are produced by an unpublished try-off pipeline and whose validation is deferred to a missing supplementary. The quantitative support also lacks error bars, significance tests, and a specified human-evaluation protocol. These gaps are load-bearing for the paper's headline claim, so the significance is currently conditional.

major comments (4)
  1. [§3.1, §4.1, Table 2] The multi-garment benchmark is evaluated only on WearWow-2K, whose target agnostic images are generated by an automated try-off pipeline. The paper states in §3.1 that dataset details are deferred to the supplementary and in §4.1 that detailed distributions and visual examples are deferred to the supplementary, which is not included. Because the same synthetic distribution is used for training and testing, systematic artifacts (smoothing, ghosting, body-shape bias) would be learned by the model and inflate the reported FID/KID/LPIPS/HE. The abstract's claim of 'physically correct' and 'flawless' agnostic models is unsupported. The authors should provide the try-off architecture, manual/automated validation of the agnostic images, and ideally an independent test set using real user try-off photos to demonstrate generalization. This is load-bearing for the central SOTA claim.
  2. [Table 1, Table 2, §4.3 (Human Evaluation)] No error bars, confidence intervals, or significance tests are reported for FID, KID, LPIPS, SSIM, or HE. KID is known to be high-variance; differences like 7.54 vs 8.18 (Table 2) may not be statistically distinguishable without repeated trials. The human evaluation protocol is underspecified: how many ratings per comparison, how participants were instructed, whether outputs were randomized and blinded, and what the variance across participants was. The statement 'a clear user preference' (HE 0.2543 vs 0.2229) needs statistical support. Without this, the 'new state-of-the-art' claim is not quantitatively established.
  3. [§3.3, Eq. (1)–(2)] The CDR preference pairs are constructed by diffusing a high-quality (chosen) and a low-quality (rejected) sample to the same noise level, but the paper does not specify how the low-quality samples are obtained or labeled. The text only says 'we curate a comparative subset based on the physical material fidelity of generated results.' If the rejected samples are, for example, SFT outputs or over-smoothed variants, this must be stated precisely. The implicit reward R_theta in Eq. (2) depends on this choice: if the reject distribution is too close to or too far from the chosen distribution, the preference signal degenerates. This is a correctness-risk issue for the MTR method and should be clarified.
  4. [§4.3, Table 2 caption] The multi-garment baselines are a mix of open local models and commercial APIs, but the comparison protocol is not specified. It is unclear what prompts, garment inputs, or reference masks were used for each baseline, and whether each baseline had equal access to the same garment images and textual descriptions. Commercial APIs may have different internal prompt parsing or resolution handling; without a controlled protocol, the comparison may be unfair. Also, reference [3] is cited as GPT-image 1.5 but points to the DALL-E 3 technical report, and [30]/[31] are duplicates; these should be corrected.
minor comments (6)
  1. [Figure 2] The figure shows 'Sparse Density: 40% ～70%' and '70% ～ 90%', while the text in §3.2 says the baseline token utilization is approximately 40% and pruning elevates it from 70% to over 90%. Please align the figure labels with the text and clarify the two stages.
  2. [§3.3, Eq. (4)] The role of ω is described as a 'relative CFG scaling coefficient,' but the equation itself is a simple margin loss. Please define how ω interacts with CFG or justify the name.
  3. [§4.2] The backbone is described as 'pre-trained Qwen-Image-Edit foundation model' with a Qwen2.5VL encoder, but no citation or details of the edit model are given. It would help to state the exact checkpoint and any conditioning interface used.
  4. [Throughout] There are several typos and formatting issues: 'T able 1', 'sparisity', 'W earW ow' in table captions, and inconsistent capitalization in figure labels. A careful proofread is needed.
  5. [§4.1] The release status of WearWow-2K is not stated (e.g., whether it will be public and under what license). This matters for reproducibility and for the community's ability to independently verify the multi-garment benchmark.
  6. [§4.4, Table 1] The discussion of SSIM favoring over-smoothed outputs is a reasonable caveat, but it is not quantified. Consider reporting a no-reference texture metric or frequency-domain analysis to substantiate the claim that the LPIPS/FID improvements reflect restored high-frequency details rather than other perceptual differences.

Circularity Check

0 steps flagged

No central circularity: ATP and MTR are self-contained; the WearWow-2K validation gap is a benchmark-validity concern, not a definitional reduction.

full rationale

The claimed derivation chain is not circular by construction. ATP is an algorithmic condition-compression mechanism; its complexity bound O((N_base+L-m)^2) is a direct consequence of self-attention scaling and the pruned token count, and it is not fitted to any reported metric. MTR is a preference-optimization post-training stage built on external, frozen scoring models (CLIP/HPSv2) and on the published DPO and SRPO formulations; the CDR/SGR losses are training objectives, not the headline evaluation metrics, and no parameter is fitted to FID/KID/LPIPS or the human evaluation. The single-garment experiments on VITON-HD and DressCode give an independent external check of the model's image quality. The multi-garment SOTA claim is measured only on the authors' WearWow-2K set, whose clothing-agnostic conditions are produced by the same try-off pipeline used for training; the paper defers the validation of those synthesized conditions to a missing supplementary. That is a genuine generalization/validation risk (the test may not transfer to real user photos), but it is not an equivalence of input and output by construction: the target images are real photographs, the baselines and WearWow are scored on identical inputs, and no test statistic is an algebraic consequence of a trainable parameter. Self-citations (CatVTON, MMTryon, etc.) appear as baselines/related work and do not carry the derivation. I therefore find no significant circularity.

Axiom & Free-Parameter Ledger

6 free parameters · 6 axioms · 0 invented entities

The framework relies on a domain assumption about spectral bias without spectral analysis; synthetic agnostic data whose physical correctness is unverified; preference pairs labeled by the authors' own criteria; an implicit reward (Eq. 2) with no derivation; and an ATP spatial-priority preservation claim that is asserted rather than proven. The single-garment external benchmarks provide some independent grounding, but the multi-garment claim rests on self-created data and unvalidated reward signals.

free parameters (6)
  • beta (CDR scale) = 0.5
    Scales reward margin in Eq. (2); set empirically in Section 4.2.
  • lambda (MTR balancing weight) = 0.65
    Weights CDR vs SGR in Eq. (5); tuned by hand.
  • tau (semantic margin) = 0.5
    Margin in ReLU loss Eq. (4); chosen empirically.
  • omega (relative CFG coefficient) = sampled in [0.1, 0.6] over timesteps
    Scales positive semantic similarity in Eq. (4); described as a time-varying schedule.
  • alpha_t (timestep-dependent CDR weight) = unspecified schedule
    Appears in Eq. (5) as damping of CDR at extreme noise; formula is not given.
  • canvas token capacity L = not specified
    ATP's fixed latent capacity in Section 3.2; no value or selection criterion is reported.
axioms (6)
  • standard math Quadratic self-attention complexity O(N^2) and VAE tokenization are assumed.
    Used in Section 3.2 to motivate ATP; standard result for transformer backbones.
  • domain assumption L2 diffusion training has a spectral bias that over-smooths high-frequency fabric at 2K.
    Stated in Sections 1 and 3.3 without spectral measurements or citation; motivates the entire MTR pipeline.
  • domain assumption Synthesized clothing-agnostic persons are physically correct and support mask-free training.
    Section 3.1/4.1; construction details and validation are deferred to the supplementary, so this is unverified.
  • ad hoc to paper CDR implicit reward R_theta in Eq. (2) is a valid preference reward for fabric fidelity.
    Defined as a difference of velocity-field errors; no derivation or validation connects this to human preference for fabric texture.
  • domain assumption HPSv2/CLIP semantic scores provide a useful reward signal for tactile restoration.
    Eq. (3) uses CLIP cosine similarities while Section 4.2 says HPSv2 is used; no validation that these scores track high-frequency texture quality.
  • ad hoc to paper Packing heterogeneous garments into one 2D canvas and pruning background tokens preserves spatial priors and prevents feature leakage.
    Section 3.2 asserts this 'rigorously'; no theoretical or empirical proof is provided.

pith-pipeline@v1.3.0-alltime-deepseek · 13252 in / 15991 out tokens · 150772 ms · 2026-08-01T11:17:44.351560+00:00 · methodology

0 comments
read the original abstract

Synthesizing native 2K multi-garment virtual try-on is a formidable frontier in digital fashion, critically bottlenecked by two fundamental limitations: the O(N^2) memory explosion induced by 2k conditions, and the spectral bias of diffusion models that over-smooths high-frequency fabric details. We present WearWow, an end-to-end, mask-free generative framework that pioneers ultra-high-resolution multi-garment synthesis. To mitigate the memory explosion , we propose Adaptive 2D Token Packing (ATP). ATP leverages inherent garment sparsity to algorithmically pack heterogeneous items onto a unified 2D canvas and prune uninformative background tokens, minimizing the effective sequence length and subsequent memory overhead while rigorously preserving 2D spatial priors. To rectify texture degradation, we introduce the Multi-dimensional Try-on Reward (MTR) system. MTR synergizes a Semantic Guidance Reward to explicitly drive tactile restoration with a Cloth Distribution Reward to implicitly anchor the physical distribution, a joint formulation that effectively mitigates the severe reward hacking. Furthermore, we curate WearWow-2K, an extreme-quality dataset comprising native 2K triplets, providing physically correct spatial interactions that naturally empower the model's mask-free generation. Extensive experiments demonstrate that WearWow establishes a new state-of-the-art, exceeding existing commercial baselines in native 2K multi-garment synthesis.

Figures

Figures reproduced from arXiv: 2607.19923 by Dongliang Shao, Jiang Li, Liping Wu, Luoqi Liu, Runyan Du, Song Chang, Wei Luo, Xiaochao Qu, Xiaodan Liang, Xujie Zhang.

Figure 1
Figure 1. Figure 1: Native 2K Multi-Garment Try-On with WearWow. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The overall framework of WearWow. Our pipeline enables mask-free, native 2K multi-garment virtual try-on through two core mechanisms. (Top) Adaptive 2D To￾ken Packing: Given a set of reference garments C, ATP algorithmically packs them onto a unified spatial canvas. Following VAE encoding, a Sparisity Token Pruning mechanism aggressively discards background void tokens. This drastically minimizes the effec… view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparisons on VITON-HD and DressCode in the single try-on. Compared with other methods, WearWow produces more texture-consistent images. MTR fine-tuning runs for an additional 800 steps at a reduced learning rate of 1×10−5 , with hyperparameters empirically set to CFG scale ω ∈ [0.1, 0.6] (sam￾pled as the timesteps changing), safety margin τ = 0.5, CDR scale β = 0.5, and balancing weight λ = 0… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparisons on WearWow-2k,showcasing our superior photorealism on complex multi-item combinations and challenging woolen textures. Fréchet Inception Distance (FID [34]). We also report Kernel Inception Distance (KID [4]) for unbiased distribution measurement and Structural Similarity Index (SSIM [39]) for spatial alignment. For ablation study, CLIP score is also com￾puted by measuring the simil… view at source ↗
Figure 5
Figure 5. Figure 5: We visualize the impact of our proposed modules. our full MTR pipeline (Ours) robustly restores micro-level tactile realism without sacrificing structural integrity. WearWow exhibits superior capability in restoring high-frequency details. No￾tably, when processing challenging fabrics like rugged denim or complex texture garment , our framework demonstrates exceptional stability and consistency. Multi-Garm… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

52 extracted references · 21 linked inside Pith

  1. [1]

    arXiv preprint arXiv:2303.08774 (2023) 5

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: Gpt-4 technical report. arXiv preprint arXiv:2303.08774 (2023) 5

    Pith/arXiv arXiv 2023

  2. [2]

    In: European Conference on Computer Vision (2022) 2

    Bai, S., Zhou, H., Li, Z., Zhou, C., Yang, H.: Single stage virtual try-on via de- formable attention flows. In: European Conference on Computer Vision (2022) 2

    2022

  3. [3]

    https://cdn

    Betker, J., Goh, G., Jing, L., Brooks, T., Wang, J., Li, L., Ouyang, L., Zhuang, J., Lee,J.,Guo,Y.,etal.:Improvingimagegenerationwithbettercaptions.Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf2(3), 8 (2023) 10, 12

    2023

  4. [4]

    Bińkowski, M., Sutherland, D.J., Arbel, M., Gretton, A.: Demystifying mmd gans (2021),https://arxiv.org/abs/1801.0140111

    Pith/arXiv arXiv 2021

  5. [5]

    arXiv preprint arXiv:2307.09481 (2023) 4

    Chen, X., Huang, L., Liu, Y., Shen, Y., Zhao, D., Zhao, H.: Anydoor: Zero-shot object-level image customization. arXiv preprint arXiv:2307.09481 (2023) 4

    Pith/arXiv arXiv 2023

  6. [6]

    In: Proc

    Choi, S., Park, S., Lee, M., Choo, J.: Viton-hd: High-resolution virtual try-on via misalignment-aware normalization. In: Proc. of the IEEE conference on computer vision and pattern recognition (CVPR) (2021) 10

    2021

  7. [7]

    arXiv preprint arXiv:2403.05139 (2024) 4, 10, 11, 12 16 X

    Choi, Y., Kwak, S., Lee, K., Choi, H., Shin, J.: Improving diffusion models for authentic virtual try-on in the wild. arXiv preprint arXiv:2403.05139 (2024) 4, 10, 11, 12 16 X. Zhang. et al

    Pith/arXiv arXiv 2024

  8. [8]

    Chong, Z., Dong, X., Li, H., Zhang, S., Zhang, W., Zhang, X., Zhao, H., Liang, X.: Catvton: Concatenation is all you need for virtual try-on with diffusion models (2024),https://arxiv.org/abs/2407.158864, 10, 11, 12

    Pith/arXiv arXiv 2024

  9. [9]

    In: Proceedings of the IEEE/CVF international conference on computer vision (2021) 4

    Cui, A., McKee, D., Lazebnik, S.: Dressing in order: Recurrent person image gen- eration for pose transfer, virtual try-on and outfit editing. In: Proceedings of the IEEE/CVF international conference on computer vision (2021) 4

    2021

  10. [10]

    In: Proceedings of the IEEE/CVF inter- national conference on computer vision (2019) 4

    Dong, H., Liang, X., Shen, X., Wang, B., Lai, H., Zhu, J., Hu, Z., Yin, J.: Towards multi-pose guided virtual try-on network. In: Proceedings of the IEEE/CVF inter- national conference on computer vision (2019) 4

    2019

  11. [11]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2022) 4

    Dong, X., Zhao, F., Xie, Z., Zhang, X., Du, D.K., Zheng, M., Long, X., Liang, X., Yang, J.: Dressing in the wild by watching dance videos. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2022) 4

    2022

  12. [12]

    arXiv preprint arXiv:2508.13632 (2025) 4, 10, 12

    Feng, Y., Zhang, L., Cao, H., Chen, Y., Feng, X., Cao, J., Wu, Y., Wang, B.: Omnitry: Virtual try-on anything without masks. arXiv preprint arXiv:2508.13632 (2025) 4, 10, 12

    Pith/arXiv arXiv 2025

  13. [13]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2021) 2

    Ge, Y., Song, Y., Zhang, R., Ge, C., Liu, W., Luo, P.: Parser-free virtual try-on via distilling appearance flows. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2021) 2

    2021

  14. [14]

    In: Proceedings of the 31st ACM International Conference on Multimedia (2023) 2

    Gou,J.,Sun,S.,Zhang,J.,Si,J.,Qian,C.,Zhang,L.:Tamingthepowerofdiffusion models for high-quality virtual try-on with appearance flow. In: Proceedings of the 31st ACM International Conference on Multimedia (2023) 2

    2023

  15. [15]

    In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision

    Guo,H.,Zeng,B.,Song,Y.,Zhang,W.,Liu,J.,Zhang,C.:Any2anytryon:Leverag- ing adaptive position embeddings for versatile virtual clothing tasks. In: Proceed- ings of the IEEE/CVF International Conference on Computer Vision. pp. 19085– 19096 (2025) 4, 10, 12

    2025

  16. [16]

    In: CVPR (2022) 4

    He, S., Song, Y.Z., Xiang, T.: Style-based global appearance flow for virtual try-on. In: CVPR (2022) 4

    2022

  17. [17]

    arXiv preprint arXiv:2208.01626 (2022) 4

    Hertz, A., Mokady, R., Tenenbaum, J., Aberman, K., Pritch, Y., Cohen-Or, D.: Prompt-to-prompt image editing with cross attention control. arXiv preprint arXiv:2208.01626 (2022) 4

    Pith/arXiv arXiv 2022

  18. [18]

    arXiv preprint arXiv:2302.09778 (2023) 4

    Huang, L., Chen, D., Liu, Y., Shen, Y., Zhao, D., Zhou, J.: Composer: Cre- ative and controllable image synthesis with composable conditions. arXiv preprint arXiv:2302.09778 (2023) 4

    Pith/arXiv arXiv 2023

  19. [19]

    Kim, J., Jin, H., Park, S., Choo, J.: Promptdresser: Improving the quality and controllability of virtual try-on via generative textual prompt and prompt-aware mask (2024),https://arxiv.org/abs/2412.1697810, 12

    Pith/arXiv arXiv 2024

  20. [20]

    In: European Conference on Com- puter Vision (2022) 4

    Lee, S., Gu, G., Park, S., Choi, S., Choo, J.: High-resolution virtual try-on with misalignment and occlusion-handled conditions. In: European Conference on Com- puter Vision (2022) 4

    2022

  21. [21]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Li, G., Wang, Y., Luan, J., Zhao, L., Xing, W., Lin, H., Ou, B.: Cascaded diffusion models for virtual try-on: Improving control and resolution. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 4689–4697 (2025) 4

    2025

  22. [22]

    In: Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)

    Lin, E., Zhang, X., Zhao, F., Luo, Y., Dong, X., Zeng, L., Liang, X.: Dreamfit: Garment-centric human generation via a lightweight anything-dressing encoder. In: Proceedings of the AAAI Conference on Artificial Intelligence (AAAI). pp. 5218–5226 (2025), aAAI 2025 4

    2025

  23. [23]

    In: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops (June 2020) 4 WearWow 17

    Minar, M.R., Tuan, T.T., Ahn, H., Rosin, P., Lai, Y.K.: Cp-vton+: Clothing shape and texture preserving image-based virtual try-on. In: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops (June 2020) 4 WearWow 17

    2020

  24. [24]

    In: Proceedings of the ACM International Conference on Multimedia (2023) 2

    Morelli, D., Baldrati, A., Cartella, G., Cornia, M., Bertini, M., Cucchiara, R.: LaDI-VTON: Latent Diffusion Textual-Inversion Enhanced Virtual Try-On. In: Proceedings of the ACM International Conference on Multimedia (2023) 2

    2023

  25. [25]

    Morelli, D., Fincato, M., Cornia, M., Landi, F., Cesari, F., Cucchiara, R.: Dress code: High-resolution multi-category virtual try-on (2022) 10

    2022

  26. [26]

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., Sutskever, I.: Learning transferable visual models from natural language supervision (2021) 8

    2021

  27. [27]

    Advances in neural information processing systems36, 53728–53741 (2023) 7

    Rafailov, R., Sharma, A., Mitchell, E., Manning, C.D., Ermon, S., Finn, C.: Direct preference optimization: Your language model is secretly a reward model. Advances in neural information processing systems36, 53728–53741 (2023) 7

    2023

  28. [28]

    arXiv preprint arXiv:2204.06125 (2022) 4

    Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., Chen, M.: Hierarchical text- conditional image generation with clip latents. arXiv preprint arXiv:2204.06125 (2022) 4

    Pith/arXiv arXiv 2022

  29. [29]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2022) 4

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2022) 4

    2022

  30. [30]

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E., Ghasemipour, S.K.S., Ayan, B.K., Mahdavi, S.S., Lopes, R.G., Salimans, T., Ho, J., Fleet, D.J., Norouzi, M.: Photorealistic text-to-image diffusion models with deep language un- derstanding (2022) 10

    2022

  31. [31]

    Advances in neural information processing systems35, 36479–36494 (2022) 12

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E.L., Ghasemipour, K., Gontijo Lopes, R., Karagol Ayan, B., Salimans, T., et al.: Photorealistic text- to-image diffusion models with deep language understanding. Advances in neural information processing systems35, 36479–36494 (2022) 12

    2022

  32. [32]

    arXiv preprint arXiv:1707.06347 (2017) 7

    Schulman, J., Wolski, F., Dhariwal, P., Radford, A., Klimov, O.: Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347 (2017) 7

    Pith/arXiv arXiv 2017

  33. [33]

    arXiv preprint arXiv:2509.20427 (2025) 5, 10, 12

    Seedream, T., Chen, Y., Gao, Y., Gong, L., Guo, M., Guo, Q., Guo, Z., Hou, X., Huang, W., Huang, Y., et al.: Seedream 4.0: Toward next-generation multimodal image generation. arXiv preprint arXiv:2509.20427 (2025) 5, 10, 12

    Pith/arXiv arXiv 2025

  34. [34]

    Seitzer, M.: pytorch-fid: FID Score for PyTorch.https://github.com/mseitzer/ pytorch-fid(August 2020), version 0.3.0 11

    2020

  35. [35]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Shen, F., Jiang, X., He, X., Ye, H., Wang, C., Du, X., Li, Z., Tang, J.: Imagdressing- v1: Customizable virtual dressing. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 6795–6804 (2025) 4, 10, 12

    2025

  36. [36]

    arXiv preprint arXiv:2509.06942 (2025) 8

    Shen, X., Li, Z., Yang, Z., Zhang, S., Zhang, Y., Li, D., Wang, C., Lu, Q., Tang, Y.: Directly aligning the full diffusion trajectory with fine-grained human preference. arXiv preprint arXiv:2509.06942 (2025) 8

    Pith/arXiv arXiv 2025

  37. [37]

    arXiv preprint arXiv:2512.16776 (2025) 5, 10, 12

    Team, K., Chen, J., Ci, Y., Du, X., Feng, Z., Gai, K., Guo, S., Han, F., He, J., He, K., et al.: Kling-omni technical report. arXiv preprint arXiv:2512.16776 (2025) 5, 10, 12

    Pith/arXiv arXiv 2025

  38. [38]

    In: Proceedings of the European Conference on Computer Vision (ECCV)

    Wang, B., Zheng, H., Liang, X., Chen, Y., Lin, L.: Toward characteristic-preserving image-based virtual try-on network. In: Proceedings of the European Conference on Computer Vision (ECCV). pp. 589–604 (2018) 4

    2018

  39. [39]

    IEEE transactions on image processing 13(4), 600–612 (2004) 11

    Wang, Z., Bovik, A.C., Sheikh, H.R., Simoncelli, E.P.: Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing 13(4), 600–612 (2004) 11

    2004

  40. [40]

    arXiv preprint arXiv:2508.02324 (2025) 5, 10, 12 18 X

    Wu, C., Li, J., Zhou, J., Lin, J., Gao, K., Yan, K., Yin, S.m., Bai, S., Xu, X., Chen, Y., et al.: Qwen-image technical report. arXiv preprint arXiv:2508.02324 (2025) 5, 10, 12 18 X. Zhang. et al

    Pith/arXiv arXiv 2025

  41. [41]

    arXiv preprint arXiv:2306.09341 (2023) 9

    Wu, X., Hao, Y., Sun, K., Chen, Y., Zhu, F., Zhao, R., Li, H.: Human preference score v2: A solid benchmark for evaluating human preferences of text-to-image synthesis. arXiv preprint arXiv:2306.09341 (2023) 9

    Pith/arXiv arXiv 2023

  42. [42]

    In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (2023) 2, 4

    Xie, Z., Huang, Z., Dong, X., Zhao, F., Dong, H., Zhang, X., Zhu, F., Liang, X.: Gp-vton: Towards general purpose virtual try-on via collaborative local-flow global-parsing learning. In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (2023) 2, 4

    2023

  43. [43]

    arXiv preprint arXiv:2207.13475 (2022) 4

    Xie, Z., Huang, Z., Zhao, F., Dong, H., Kampffmeyer, M., Dong, X., Zhu, F., Liang, X.: Pasta-gan++: A versatile framework for high-resolution unpaired virtual try- on. arXiv preprint arXiv:2207.13475 (2022) 4

    Pith/arXiv arXiv 2022

  44. [44]

    In: Proceedings of the 29th ACM International Conference on Multimedia (2021) 4

    Xie, Z., Zhang, X., Zhao, F., Dong, H., Kampffmeyer, M.C., Yan, H., Liang, X.: Was-vton: Warping architecture search for virtual try-on network. In: Proceedings of the 29th ACM International Conference on Multimedia (2021) 4

    2021

  45. [45]

    arXiv preprint arXiv:2403.01779 (2024) 4, 10, 12

    Xu, Y., Gu, T., Chen, W., Chen, C.: Ootdiffusion: Outfitting fusion based latent diffusion for controllable virtual try-on. arXiv preprint arXiv:2403.01779 (2024) 4, 10, 12

    Pith/arXiv arXiv 2024

  46. [46]

    arXiv preprint arXiv:2505.09388 (2025) 9

    Yang, A., Li, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Gao, C., Huang, C., Lv, C., et al.: Qwen3 technical report. arXiv preprint arXiv:2505.09388 (2025) 9

    Pith/arXiv arXiv 2025

  47. [47]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023) 4

    Yang, B., Gu, S., Zhang, B., Zhang, T., Chen, X., Sun, X., Chen, D., Wen, F.: Paint by example: Exemplar-based image editing with diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023) 4

    2023

  48. [48]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2020) 4

    Yang, H., Zhang, R., Guo, X., Liu, W., Zuo, W., Luo, P.: Towards photo-realistic virtual try-on by adaptively generating-preserving image content. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (2020) 4

    2020

  49. [49]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zhang, R., Isola, P., Efros, A.A., Shechtman, E., Wang, O.: The unreasonable effectiveness of deep features as a perceptual metric. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 586–595 (2018) 10

    2018

  50. [50]

    Zhang, X., Lin, E., Li, X., Luo, Y., Kampffmeyer, M., Dong, X., Liang, X.: Mmtryon: Multi-modal multi-reference control for high-quality fashion generation (2024),https://arxiv.org/abs/2405.004484

    Pith/arXiv arXiv 2024

  51. [51]

    arXiv preprint arXiv:2412.08486 (2024) 4, 10, 11, 12

    Zhou, Z., Liu, S., Han, X., Liu, H., Ng, K.W., Xie, T., Cong, Y., Li, H., Xu, M., Pérez-Rúa, J.M., Patel, A., Xiang, T., Shi, M., He, S.: Learning flow fields in at- tention for controllable person image generation. arXiv preprint arXiv:2412.08486 (2024) 4, 10, 11, 12

    Pith/arXiv arXiv 2024

  52. [52]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023) 4

    Zhu, L., Yang, D., Zhu, T., Reda, F., Chan, W., Saharia, C., Norouzi, M., Kemelmacher-Shlizerman, I.: Tryondiffusion: A tale of two unets. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023) 4

    2023