DuoMatching: Joint-Marginal Distribution Matching for Few-Step Video Generation

DuoMatching improves few-step video generation by jointly matching frame distributions and adding frame-level supervision via an image teacher, boosting visual quality and semantic alignment over 80%.

Jiahao Zhan, Yan Wang, Yongrui Ma, Qunliang Xing, Ruchang Yao, Runtao Liu, Shijie Zhao, Tianfan Xue

Published Oct 2, 2026▲ 17 on Hugging Face Code ★ 4 arXiv ↗

78%

OverallHighly rated

OverallHighly ratedVote to see the scoreThe exact score shows once you've voted, so every vote is your own call. The first half of each home page shelf shows its scores.

Readers

?1 reader voted. Vote to see how they split.

Only vote on papers you've read. Sign in with GitHub to vote.

AI panel9/20reviewers recommend it

lenient 5/5

medium 4/10

strict 0/5

AI panel?Vote to see what the 20 AI reviewers said

Panel consensus

DuoMatching's joint-marginal distillation and LatentBridge improve visual quality and semantic alignment with strong human preference, though it lacks temporal consistency metrics, open-source artifacts, and proof that gains come from distribution matching rather than image-prior smoothing.

Abstract

Streaming video generation has benefited from distribution matching distillation (DMD), which matches the joint distribution of video frames to a video teacher's approximation of the real video distribution. Although this joint matching mitigates drift during autoregressive rollouts, limitations remain in visual quality and semantic alignment. To address these limitations, we propose DuoMatching, a distribution matching framework that approximates the real video distribution through a unified joint-marginal formulation. On top of existing joint matching formulations, the additional marginal matching objective provides dedicated frame-level supervision from an image generator, transferring complementary visual and semantic priors from it. To apply this frame-level supervision in video generation, we introduce LatentBridge to resolve the latent representation mismatch between the video student and the image teacher. Latent Variation Sampling further distributes such frame-level supervision across distinct temporal segments, reducing redundancy. Experiments demonstrate that DuoMatching improves visual quality, composition, and semantic alignment while largely preserving motion dynamics. Human evaluations show overall preference rates above 80% against all evaluated baselines. The project page is available at https://johnzhan2023.github.io/DuoMatching/.