AI Paper Atlas

Research map / Video / Video generation

Nexus-Vid: Efficient Frequency Bridging with Homogeneous Latent Space for Video Unified Models

Jiazheng Xing, Hangjie Yuan, Lingling Cai, Xinyu Liu, Yujie Wei, Fei Du, Tao Feng, Hai Ci, Jiasheng Tang, Weihua Chen, Fan Wang, Yong Liu

ECCV 2026 · Video › Video generation · approach: Autoregressive long video generation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Hand off to a large Wan2.1-14B generator at inference via UPFBOmni-Video · VBench-T2VTotal score: 83.82 -> 84.12Improved
Same change, semantic scoreOmni-Video · VBench-T2VSemantic score: 79.10 -> 80.52Improved
Same change on reasoning benchmarkOmni-Video · VR-BenchTotal score: 72.78 -> 79.28Improved
Swap large generator to Wan2.2-A14BLumos-Nexus · VR-BenchTotal score: 80.98 -> 81.90Improved
Use temporal gating sharpness gamma_w 0.3 rather than 0.5 (ablation)UPFB · VR-BenchTotal score: 75.10 -> 79.28Improved
Vary gamma_w on VBench total (ablation)UPFB · VBench-T2VTotal score: 84.02 to 84.12No clear effect
Use moderate bandwidth schedule (0.35, 0.70) instead of wide (1.00, 2.00) (ablation)UPFB · VBench-T2VTotal score: 83.56 -> 84.12Improved
Use moderate bandwidth schedule instead of narrow (0.05, 0.10) (ablation)UPFB · VBench-T2VTotal score: 83.99 -> 84.12Improved
Add RMS alignment and energy re-balancing (ablation)UPFB · VBench-T2VTotal score: 84.07 -> 84.12Improved
Use frequency-aware progressive bridging instead of Direct Add averaging (ablation)Lumos-Nexus · VBench-T2VTotal score: 83.28 -> 84.12Improved
Same, on VR-Bench (ablation)Lumos-Nexus · VR-BenchTotal score: 77.33 -> 79.28Improved
Run both generators at inferenceWan2.1-14B · Inference costLatency: 35.1 -> 40.5 s/stepHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy