Paper page - FashionChameleon: Towards Real-Time and Interactive Human-Garment Video Customization

Source: https://huggingface.co/papers/2605.15824

Abstract

FashionChameleon enables real-time interactive multi-garment video customization through teacher-student distillation and in-context learning techniques while maintaining motion coherence.

Human-centric video customization, particularly at the garment level, has shown significant commercial value. However, existing approaches cannot support low-latency and interactive garment control, which is crucial for applications such as e-commerce and content creation. This paper studies how to achieve interactive multi-garment video customization while preservingmotion coherenceusing onlysingle-garment video data. We present FashionChameleon, a real-time and interactive framework for human-garment customization inautoregressive video generation, where users can interactively switch garment during generation. FashionChameleon consists of three key techniques: (i) Instead of training on multi-garment video data, we train a Teacher Model withIn-Context Learningon a single reference-garment pair. By retaining the image-to-video training paradigm while enforcing a mismatch between the reference and garment image, the model is encouraged to implicitly preserve coherence during single-garment switching. (ii) To achieve consistency and efficiency during generation, we introduceStreaming DistillationwithIn-Context Learning, which fine-tunes the model with in-context teacher forcing and improves extrapolation consistency viagradient-reweighted distribution matchingdistillation. (iii) To extend the model for interactive multi-garment video customization, we propose Training-FreeKV Cache Rescheduling, which includes garment KV refresh, historical KV withdraw, and reference KV disentangle to achievegarment switchingwhile preservingmotion coherence. Our FashionChameleon uniquely supports interactive customization and consistent long-video extrapolation, while achievingreal-time generationat 23.8 FPS on a single GPU, 30-180times faster than existing baselines.

View arXiv pageView PDFProject pageGitHub4Add to collection

Get this paper in your agent:

hf papers read 2605\.15824

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/2605.15824 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.15824 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.15824 in a Space README.md to link it from this page.

Collections including this paper1