TODO: Kling 3.0 Text-to-Video enrichment prompt.
Model constraints: 75-100 words, balanced scene + action description.