ResearchBenchmarksText-to-video generation on VBench (Detailed Fidelity, Consistency, and Quality Metrics)Follow82Object Class FidelityPFM76.07277.61179.1580.689Jul 3, 2026Evaluation ResultsMethodMethodLinksObject Class FidelityMulti-Object FidelityHuman Action FidelityColor FidelitySpatial Relationship FidelityScene FidelityAppearance Style ScoreTemporal Style ScoreOverall Consistency ScoreSubject Consistency ScoreBackground Consistency ScoreTemporal Flicker ScoreMotion Smoothness ScoreDynamic Degree ScoreAesthetic Quality ScoreImaging Quality ScoreOverall Quality ScoreSemantic ScoreTotal ScorePFMNFE=8NFE=82026.078256.29178.96235.619.821.424.696.997.299.699.531.965.161.88267.779.2Wan 2.1NFE=35×2NFE=35×22026.0778.659.28191.264.628.322.221.921.996.29697.998.737.953.565.879.967.577.4Wan 2.1NFE=8×2NFE=8×22026.0776.338.38067.935.12622.618.622.291.896.499.497.651.45546.577.557.873.5