Oxygen-TryOn: Fashion-Native Foundation Model for Any-item Virtual Try-On

Hugging Face Daily Papers Papers
virtual-try-on foundation-model fashion-ai image-generation multi-reference reinforcement-learning

Summary

Oxygen-TryOn is a unified foundation model for any-item virtual try-on, achieving state-of-the-art consistency and realism across single and multi-item try-on tasks through a dedicated data engine and three-stage training pipeline.

We present Oxygen-TryOn, a unified foundation model for any-item virtual try-on. Rather than repurposing a general-purpose image editor, Oxygen-TryOn is fashion-native, built for try-on through a dedicated data engine and try-on-specific training. Given one or more reference items (clean product shots or in-the-wild worn-on photos) and a single target subject image, it synthesizes a photorealistic image of the subject wearing the items across virtually any fashion category. Prior systems handle a single garment category in a studio setting, and recent multi-reference methods remain garment-centric; in contrast, Oxygen-TryOn supports diverse items and scenarios, including full- and half-body views, a variable number of references, and free multi-item composition, while faithfully preserving both subject identity and item appearance. Instead of mask-based inpainting, we reformulate try-on as a multi-reference, understanding-driven generation task. We build a data engine that collects, manufactures, annotates, and filters high-quality try-on data at scale, and design a three-stage recipe of continued pre-training (CPT), supervised fine-tuning (SFT), and reinforcement learning (RL). The RL stage uses a hybrid reward combining an in-house try-on reward model with a proprietary, rubric-guided general-purpose model, jointly supervising fine-grained consistency and instruction-level quality. It also follows general editing instructions (e.g., pose changes) in the same pass. Across public benchmarks and our in-house Oxygen-TryOn Bench, it achieves state-of-the-art consistency and realism on single-item try-on and leads on multi-item try-on, matching or surpassing both leading proprietary systems (Nano Banana Pro, GPT-Image-2, Seedream5 Lite) and open-source models (FLUX.2).
Original Article
View Cached Full Text

Cached at: 07/28/26, 06:33 AM

Paper page - Oxygen-TryOn: Fashion-Native Foundation Model for Any-item Virtual Try-On

Source: https://huggingface.co/papers/2607.21694 Authors:

,

,

,

,

,

,

,

,

,

,

,

Abstract

WepresentOxygen-TryOn,aunifiedfoundationmodelforany-itemvirtualtry-on.Ratherthanrepurposingageneral-purposeimageeditor,Oxygen-TryOnisfashion-native,builtfortry-onthroughadedicateddataengineandtry-on-specifictraining.Givenoneormorereferenceitems(cleanproductshotsorin-the-wildworn-onphotos)andasingletargetsubjectimage,itsynthesizesaphotorealisticimageofthesubjectwearingtheitemsacrossvirtuallyanyfashioncategory.Priorsystemshandleasinglegarmentcategoryinastudiosetting,andrecentmulti-referencemethodsremaingarment-centric;incontrast,Oxygen-TryOnsupportsdiverseitemsandscenarios,includingfull-andhalf-bodyviews,avariablenumberofreferences,andfreemulti-itemcomposition,whilefaithfullypreservingbothsubjectidentityanditemappearance.Insteadofmask-basedinpainting,wereformulatetry-onasamulti-reference,understanding-drivengenerationtask.Webuildadataenginethatcollects,manufactures,annotates,andfiltershigh-qualitytry-ondataatscale,anddesignathree-stagerecipeofcontinuedpre-training(CPT),supervisedfine-tuning(SFT),andreinforcementlearning(RL).TheRLstageusesahybridrewardcombininganin-housetry-onrewardmodelwithaproprietary,rubric-guidedgeneral-purposemodel,jointlysupervisingfine-grainedconsistencyandinstruction-levelquality.Italsofollowsgeneraleditinginstructions(e.g.,posechanges)inthesamepass.Acrosspublicbenchmarksandourin-houseOxygen-TryOnBench,itachievesstate-of-the-artconsistencyandrealismonsingle-itemtry-onandleadsonmulti-itemtry-on,matchingorsurpassingbothleadingproprietarysystems(NanoBananaPro,GPT-Image-2,Seedream5Lite)andopen-sourcemodels(FLUX.2).

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2607\.21694

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.21694 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.21694 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.21694 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

CtrlVTON: Controllable Virtual Try-On via Visual-Instance-Prompt Segmentation

Hugging Face Daily Papers

This paper introduces VIP-SAM for instance-level garment segmentation and CtrlVTON, a controllable virtual try-on framework that treats try-on as an image editing problem, allowing precise control over garment layout, style, and placement. Both methods achieve state-of-the-art results on their respective tasks.

Where to Look: Can Foundation Models Reach a Target Viewpoint Through Active Exploration?

Hugging Face Daily Papers

Introduces Target Viewpoint Reproduction (TVR) task and TVRBench benchmark for evaluating foundation models' ability to actively adjust 3D viewpoints to match target images. Experiments reveal significant limitations in current open and closed-source models, with a unified post-training framework boosting success rates from ~12% to ~51%.

OpenMHC: Accelerating the Science of Wearable Foundation Models

arXiv cs.LG

OpenMHC introduces the largest open-access wearable health dataset with over 60 million hours of data and open-source implementations of wearable foundation models, including a unified benchmark for prediction, imputation, and forecasting.