AI Paper Atlas

Research map / Video / Video generation

TempoMaster: Efficient Long Video Generation via Next-Frame-Rate Prediction

Yukuo Ma, Cong Liu, Junke Wang, Junqi Liu, Haibin Huang, Zuxuan Wu, Chi Zhang, Xuelong Li

CVPR 2026 · Video › Video generation · approach: Autoregressive long video generation · 3 citations (OpenAlex)

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
generate video as next-frame-rate prediction (coarse-to-fine in frame rate) instead of autoregressive or bidirectional generationTempoMaster (Wan2.2 14B) · VBench long video (500 frames)Total score: 80.30 vs best baseline 79.52Improved
use next-frame-rate prediction on short videosTempoMaster (Wan2.2 14B) · VBench short video (121 frames)Total score: 80.76 vs best baseline 80.55Improved
compare against autoregressive-like methods on long vs short videosMAGI-1 and SkyReels-V2 · VBenchTotal score: MAGI-1 79.05 -> 78.50; SkyReels-V2 80.54 -> 79.17Hurt
VBench consistency and smoothness metrics reward low-motion videosFramePack · VBench vs human studyMotion quality: Dynamic degree 16.82 on VBench; human motion quality 2.88Mixed
TempoMaster vs baselines in human evaluationTempoMaster · Human study, long videosTotal score: 3.69 vs LongCat 3.58Improved
use two stages (6, 24 fps) with more parallel segments (1, 8) instead of (1, 4) (ablation)TempoMaster inference · VBench long video (500 frames)PFLOPs / total score: 526.48 -> 400.19 PFLOPs; 80.30 -> 80.29Improved
use parallel segments (1, 4, 8) instead of (1, 2, 4) in three stages (ablation)TempoMaster inference · VBench long video (500 frames)PFLOPs / total score: 774.23 -> 460.02 PFLOPs; 80.30 -> 80.30Improved
use parallel segments (1, 8, 8) instead of (1, 2, 4) in three stages (ablation)TempoMaster inference · VBench long video (500 frames)Total score: 80.30 -> 80.04Hurt
drop the intermediate 12 fps stage (two stages 6, 24 fps with segments (1, 4)) (ablation)TempoMaster inference · VBench short video (121 frames)Total score: 80.76 -> 80.55Hurt
use more aggressive parallelism (1, 4, 8) in three stages (ablation)TempoMaster inference · VBench short video (121 frames)Total score: 80.76 -> 80.26Hurt
add randomized temporal position indices during training (ablation)TempoMaster · VBench long video (500 frames)Total score: 80.00 -> 80.19Improved

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