Multi Dimensional Video Generation Quality Evaluation With Decomposed Metrics

via “multi-dimensional video generation quality evaluation with decomposed metrics”

[CVPR2024 Highlight] VBench - We Evaluate Video Generation

Unique: Decomposes video generation evaluation into 16-18 independent dimensions with human-preference validation, rather than single holistic scores. Uses specialized pretrained models per dimension (optical flow for motion, CLIP for semantics, action recognition for temporal understanding) and aggregates with learned weighting from human annotations. VBench-2.0 extends this with intrinsic faithfulness dimensions that measure alignment between prompts and generated content.

More interpretable than single-metric benchmarks (LPIPS, FVD) because dimension-level scores pinpoint specific quality gaps; more reproducible than human evaluation because automated metrics are deterministic and standardized across models.