Prompt Enhancement for Text-to-Video Generation on CogVideoX-5B (test)
52.6SAGPT-4o
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4oEnhanced Model=GPT-4o2026.03 | 52.6 | 66.4 | 41.8 | |
| PhyT2VEnhanced Model=PhyT2V2026.03 | 50.6 | 65 | 40.8 | |
| DeepSeek-V3Enhanced Model=DeepSeek-V32026.03 | 50.6 | 63.6 | 40.8 | |
| Original PromptEnhanced Model=N/A2026.03 | 50.2 | 63.4 | 39.4 | |
| PhyPromptModel Size=7B2026.03 | 50.2 | 69.2 | 42 | |
| Qwen2.5Model Size=7B2026.03 | 47.8 | 63.8 | 38.2 | |
| PhyPromptModel Size=1.5B2026.03 | 47.4 | 66.6 | 39 | |
| PromptistEnhanced Model=Promptist2026.03 | 46.4 | 63 | 37.4 | |
| PhyPromptModel Size=3B2026.03 | 46.4 | 68.4 | 40.2 | |
| Qwen2.5Model Size=3B2026.03 | 45.2 | 64.6 | 35.8 | |
| Qwen2.5Model Size=1.5B2026.03 | 39.2 | 66.4 | 31 |