6/ Future work: Multimodal2Effect. Support both effect descriptions and visual video examples as input, so users can specify an effect by text, example video, or both. More videos, paper, code: