# Video Generation on VBench (8 Perceptual Metrics Subset) Source: [Video Generation on VBench (8 Perceptual Metrics Subset)]() Collection index: [llms.txt]() Data updated: 2026-03-20T09:58:21.472Z Evaluation of Video Generation models on the VBench dataset, reporting scores for 8 specific perceptual metrics including Aesthetic Quality, Consistency measures (Subject, Background, Overall, Scene), Dynamic Degree, Imaging Quality, and Motion Smoothness. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Video Generation | | Dataset (including any stated split) | VBench | | Metric | Aesthetic Quality | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Aesthetic Quality](): Higher is better; unit: PERCENT. - [Background Consistency](): Higher is better; unit: PERCENT. - [Overall Consistency](): Higher is better; unit: PERCENT. - [Dynamic Degree](): Higher is better; unit: PERCENT. - [Subject Consistency](): Higher is better; unit: PERCENT. - [Imaging Quality](): Higher is better; unit: PERCENT. - [Scene Consistency](): Higher is better; unit: PERCENT. - [Motion Smoothness](): Higher is better; unit: PERCENT. ## Recorded results | Method | Aesthetic Quality | Result date | Report paper | | --- | --- | --- | --- | | Original | 59.22 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | Ours (DMPQ) | 57.24 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | Original | 54.64 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | Ours (DMPQ) | 54.37 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | QuaRot | 53.71 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | QuaRot | 53.45 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | SmoothQuant | 53.32 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | ViDiT-Q | 53.32 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | QuaRot | 53 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | QuaRot | 51.58 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | SmoothQuant | 51.48 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | SmoothQuant | 50.1 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | SmoothQuant | 49.5 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | ViDiT-Q | 47.45 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | ViDiT-Q | 45.25 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | | ViDiT-Q | 44.33 | Not recorded | 69bcd4a00636e5bf4afc2ad6 | ### Original Aesthetic Quality: 59.22. Result ID: 69bd1a2747f34ce924a7b4ef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W16/A16","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### Ours (DMPQ) Aesthetic Quality: 57.24. Result ID: 69bd1a2747f34ce924a7b4f6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### Original Aesthetic Quality: 54.64. Result ID: 69bd1a2747f34ce924a7b4e7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W16/A16","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### Ours (DMPQ) Aesthetic Quality: 54.37. Result ID: 69bd1a2747f34ce924a7b4ee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### QuaRot Aesthetic Quality: 53.71. Result ID: 69bd1a2747f34ce924a7b4f1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### QuaRot Aesthetic Quality: 53.45. Result ID: 69bd1a2747f34ce924a7b4f4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### SmoothQuant Aesthetic Quality: 53.32. Result ID: 69bd1a2747f34ce924a7b4e8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### ViDiT-Q Aesthetic Quality: 53.32. Result ID: 69bd1a2747f34ce924a7b4ea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### QuaRot Aesthetic Quality: 53. Result ID: 69bd1a2747f34ce924a7b4ec. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### QuaRot Aesthetic Quality: 51.58. Result ID: 69bd1a2747f34ce924a7b4e9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### SmoothQuant Aesthetic Quality: 51.48. Result ID: 69bd1a2747f34ce924a7b4f0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### SmoothQuant Aesthetic Quality: 50.1. Result ID: 69bd1a2747f34ce924a7b4eb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### SmoothQuant Aesthetic Quality: 49.5. Result ID: 69bd1a2747f34ce924a7b4f3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### ViDiT-Q Aesthetic Quality: 47.45. Result ID: 69bd1a2747f34ce924a7b4ed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-2B"}. This record does not assert independent reproduction. ### ViDiT-Q Aesthetic Quality: 45.25. Result ID: 69bd1a2747f34ce924a7b4f2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A8","Model":"CogVideo-5B"}. This record does not assert independent reproduction. ### ViDiT-Q Aesthetic Quality: 44.33. Result ID: 69bd1a2747f34ce924a7b4f5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Bit-width (W/A)":"W4A6","Model":"CogVideo-5B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.