Paper page - From Short Clips to Real-Time Streams: Identity-Aware Supervision and Context-Efficient Generation for Video Face Swapping

Source: https://huggingface.co/papers/2610.26681 Authors:

,

,

,

,

,

,

,

,

,

Abstract

StreamingSwap is a real-time flow-matching framework for video face swapping that improves identity fidelity and efficiency via joint identity-reconstruction loss and reduced spatial-temporal generation, achieving stable 20 FPS performance on long videos.

Video face swapping transfers the reference identity to a target video while preserving its other attributes and motion. Despite recent progress inreference-guided flow-based methods, jointly achieving faithful identity transfer, high fidelity, and real-time generation over long video streams remains challenging. To address this challenge, we rethink the task-specific structure of reference-guided video face swapping. First, face swapping changes only identity while preserving most visual content, yetflow-matchingsupervision is weakly sensitive to these changes. We therefore developJoint Rec-ID Loss, which balances explicit identity supervision and latent reconstruction according to the noise level. Second, the target video already provides non-edited attributes and temporal dynamics, enabling generation with fewer spatial tokens and shorter temporal contexts throughFace-RoI GenerationandShort-Context Rollout Training. By integrating these designs, we proposeStreamingSwap, a reference-guidedflow-matchingframework for real-time streaming video face swapping. We also introduceHumanFaceVid, comprising 1,000 hours of cinematic training data and evaluation sets of 200 short and 200 long videos. Experiments demonstrate that our method achieves competitive identity transfer and visual quality at 20 FPS on a single NVIDIA RTX 4090 GPU, with stable generation over videos up to 10 minutes long. Project webpage: https://aim-uofa.github.io/StreamingSwap/

View arXiv pageView PDFProject pageGitHub11Add to collection

Get this paper in your agent:

hf papers read 2610\.26681

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/2610.26681 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2610.26681 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2610.26681 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.