Research map / Image generation / Editing and style transfer
Virtual try-on
Warp and synthesize garments onto person images with flow and diffusion models.
46 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Editing and style transfer, keyword cluster “translation · style transfer · image image”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Still moving The best reported result is still improving on 4 of 5 tracked benchmarks.
12 papers in 2023–24 and 21 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 4.4% to 5.3%.
Open questions the data points to
- VITON-HD (SSIM): best went from 0.9 (up to 2024) to 0.911 in 2025 — Incorporating Visual Correspondence into Diffusion Model for Virtual Try-On (ICLR 2025 · 0 citations)
- VITON-HD (FID): best went from 8.54 (up to 2024) to 6.062 in 2025 — Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · 3 citations)
- VITON-HD (LPIPS): best went from 0.07 (up to 2024) to 0.061 in 2025 — Latent Diffusion-Enhanced Virtual Try-On via Optimized Pseudo-Label Generation (AAAI 2025 · 5 citations)
- VITON (SSIM): best went from 0.91 (up to 2024) to 0.93 in 2025 — GarFast: Realistic and Fast Garment Transfer with a Simplified Parser-Free Approach (AAAI 2025 · 1 citation)
- Papers report PF-AFN on VITON (FPS) anywhere from 34 to 126: comparisons against it depend on whose number you take.
- Papers report ClothFlow on VITON (FID) anywhere from 10.73 to 14.43: comparisons against it depend on whose number you take.
What looks settled
- VITON (FID): best up to 2024 8.89, best in 2025–26 9.39 from 2 papers, at most 0.2% better, if at all.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Learning to Appreciate the Aesthetic Effects of Clothing (AAAI 2016 · 23 citations)
- 2018 · VITON: An Image-Based Virtual Try-On Network (CVPR 2018 · 673 citations)
- 2020 · Towards Photo-Realistic Virtual Try-On by Adaptively Generating-Preserving Image Content (CVPR 2020)
- 2021 · VITON-HD: High-Resolution Virtual Try-On via Misalignment-Aware Normalization (CVPR 2021 · 304 citations)
- 2022 · Style-Based Global Appearance Flow for Virtual Try-On (CVPR 2022 · 122 citations)
- 2023 · GP-VTON: Towards General Purpose Virtual Try-On via Collaborative Local-Flow Global-Parsing Learning (CVPR 2023 · 97 citations)
- 2024 · CAT-DM: Controllable Accelerated Virtual Try-on with Diffusion Model (CVPR 2024 · 43 citations)
- 2025 · OOTDiffusion: Outfitting Fusion Based Latent Diffusion for Controllable Virtual Try-On (AAAI 2025 · 84 citations)
- 2026 · Inverse Virtual Try-On: Generating Multi-Category Product-Style Images from Clothed Individuals (ICLR 2026 · 0 citations)
Earliest papers here
- Learning to Appreciate the Aesthetic Effects of Clothing (AAAI 2016 · 23 citations)
- VITON: An Image-Based Virtual Try-On Network (CVPR 2018 · 673 citations)
- Towards Photo-Realistic Virtual Try-On by Adaptively Generating-Preserving Image Content (CVPR 2020)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
VITON-HD · SSIM (higher is better)
9 papers, 2021–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 0.81 | Disentangled Cycle Consistency for Highly-Realistic Virtual Try-On (CVPR 2021 · 103 citations) |
| 2023 | 0.894 | GP-VTON: Towards General Purpose Virtual Try-On via Collaborative Local-Flow Global-Parsing Learning (CVPR 2023 · 97 citations) |
| 2024 | 0.9 | Texture-Preserving Diffusion Models for High-Fidelity Virtual Try-On (CVPR 2024 · 35 citations) |
| 2025 | 0.911 | Incorporating Visual Correspondence into Diffusion Model for Virtual Try-On (ICLR 2025 · 0 citations) |
VITON-HD · FID (lower is better)
8 papers, 2021–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 15.55 | Disentangled Cycle Consistency for Highly-Realistic Virtual Try-On (CVPR 2021 · 103 citations) |
| 2023 | 9.197 | GP-VTON: Towards General Purpose Virtual Try-On via Collaborative Local-Flow Global-Parsing Learning (CVPR 2023 · 97 citations) |
| 2024 | 8.54 | Texture-Preserving Diffusion Models for High-Fidelity Virtual Try-On (CVPR 2024 · 35 citations) |
| 2025 | 6.062 | Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · 3 citations) |
VITON-HD · LPIPS (lower is better)
7 papers, 2023–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 0.0799 | GP-VTON: Towards General Purpose Virtual Try-On via Collaborative Local-Flow Global-Parsing Learning (CVPR 2023 · 97 citations) |
| 2024 | 0.07 | Texture-Preserving Diffusion Models for High-Fidelity Virtual Try-On (CVPR 2024 · 35 citations) |
| 2025 | 0.061 | Latent Diffusion-Enhanced Virtual Try-On via Optimized Pseudo-Label Generation (AAAI 2025 · 5 citations) |
VITON · FID (lower is better)
6 papers, 2021–2025 · flat since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 10.09 | Parser-Free Virtual Try-On via Distilling Appearance Flows (CVPR 2021 · 210 citations) |
| 2022 | 8.89 | Style-Based Global Appearance Flow for Virtual Try-On (CVPR 2022 · 122 citations) |
VITON · SSIM (higher is better)
5 papers, 2021–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 0.83 | Disentangled Cycle Consistency for Highly-Realistic Virtual Try-On (CVPR 2021 · 103 citations) |
| 2022 | 0.91 | Style-Based Global Appearance Flow for Virtual Try-On (CVPR 2022 · 122 citations) |
| 2025 | 0.93 | GarFast: Realistic and Fast Garment Transfer with a Simplified Parser-Free Approach (AAAI 2025 · 1 citation) |
What has been tried
33 papers here report 316 measured changes (a component added, removed or swapped, compared with a baseline): helped 82%, no clear effect 3%, hurt 14%, mixed 1%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| add outfitting dropout and use guidance scale 1.5 instead of no dropout | VITON-HD · LPIPS | 0.0750 -> 0.0705 | Helped | OOTDiffusion: Outfitting Fusion Based Latent Diffusion for Controllable Virtual Try-On (AAAI 2025 · 84 citations) |
| add image encoder branch (IEB) | Virtual dressing · ImageReward | -0.245 -> -0.178 | Helped | IMAGDressing-v1: Customizable Virtual Dressing (AAAI 2025 · 78 citations) |
| Use DINO-V2 instead of CLIP for garment feature extraction | VITON-HD unpaired · FIDu | 10.21 -> 9.67 | Helped | CAT-DM: Controllable Accelerated Virtual Try-on with Diffusion Model (CVPR 2024 · 43 citations) |
| use self-attention texture transfer instead of CLIP image features and cross-attention | VITON-HD · SSIM | 0.85 -> 0.90 | Helped | Texture-Preserving Diffusion Models for High-Fidelity Virtual Try-On (CVPR 2024 · 35 citations) |
| replace progressive single-stage training with a cascaded base and super-resolution model | Our test set · garment detail preservation | Hurt | M&M VTO: Multi-Garment Virtual Try-On and Editing (CVPR 2024 · 22 citations) | |
| add a composition mask | VITON-HD test pairs · FID | 10.91 -> 10.43 | Helped | Towards Squeezing-Averse Virtual Try-On via Sequential Deformation (AAAI 2024 · 21 citations) |
| use hierarchical CLIP features from all 8 categories instead of a single CLIP feature | Garment texture transfer · FID | 24.57 -> 22.25 | Helped | PICTURE: PhotorealistIC virtual Try-on from UnconstRained dEsigns (CVPR 2024 · 15 citations) |
| replace the mask-based pipeline with mask-free distillation | WildVTON · FIDu | 38.77 -> 57.52 | Hurt | BooW-VTON: Boosting In-the-Wild Virtual Try-On via Mask-Free Pseudo Data Training (CVPR 2025 · 13 citations) |
| add clothing pre-alignment | VITON-HD · FID | 10.984 -> 9.313 | Helped | Latent Diffusion-Enhanced Virtual Try-On via Optimized Pseudo-Label Generation (AAAI 2025 · 5 citations) |
| add pose-aware mask refinement | Ornament try-on test set · CLIP Score | 88.5 -> 88.7 | Helped | Shining Yourself: High-Fidelity Ornaments Virtual Try-on with Diffusion Model (CVPR 2025 · 5 citations) |
| add Openpose and Densepose control branches and MIM | VITON-HD · SSIM | 0.802 -> 0.896 | Helped | Cascaded Diffusion Models for Virtual Try-On: Improving Control and Resolution (AAAI 2025 · 3 citations) |
| add H2G-UH synthetic pairs during CAT-DM training | VITON-HD · FID | 8.56 -> 8.36 | Helped | Enhancing Virtual Try-On with Synthetic Pairs and Error-Aware Noise Scheduling (CVPR 2025 · 2 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| PF-AFN | VITON | FPS | 34 – 126 | 2 papers, e.g. AAAI 2024 and AAAI 2025 |
| ClothFlow | VITON | FID | 10.73 – 14.43 | 2 papers, e.g. CVPR 2022 and CVPR 2021 |
| DCI-VTON | VITON-HD | LPIPS | 0.072 – 0.08 | 2 papers, e.g. NeurIPS 2024 and AAAI 2025 |
| PF-AFN | VITON-HD | LPIPS | 0.08 – 0.0871 | 2 papers, e.g. CVPR 2024 and CVPR 2023 |
What reviewers will expect
Datasets and baselines named in the 12 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- VITON-HD · 8 of 12 papers
- DressCode · 6 of 12 papers
- VITON · 2 of 12 papers
Baselines
- LaDI-VTON · 6 of 12 papers
- IDM-VTON · 6 of 12 papers
- GP-VTON · 6 of 12 papers
- StableVITON · 5 of 12 papers
- HR-VITON · 4 of 12 papers
- DCI-VTON · 4 of 12 papers
- Paint-by-Example · 3 of 12 papers
- CAT-DM · 3 of 12 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 33 papers; from OpenAlex and the papers' affiliations).
- Wuhan University of Technology · 5 papers
- Wuhan Textile University · 3 papers
- Shanghai Artificial Intelligence Laboratory · 3 papers
- Interdisciplinary Artificial Intelligence Research Institute · 2 papers
- Tencent · 2 papers
- Shanghai Jiao Tong University · 2 papers
Use it
Download BibTeX of 32 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Image-to-image translation (127 papers)
- Texture and sketch synthesis (130 papers)
- Arbitrary neural style transfer (124 papers)
- Pose-guided person image synthesis (29 papers)
Elsewhere in the topic
- Instruction-based diffusion editing (151 papers)
- Inversion-based diffusion editing (113 papers)
- Identity-preserving face manipulation (88 papers)
- Localized and attribute-specific editing (73 papers)
- Diffusion face generation and editing (61 papers)
- GAN inversion for editing (25 papers)
- Talking face generation (21 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).