Where do you think current video AI breaks down the most? • Generation (visual quality)?• Coherence (scene flow)?• Understanding (story, intent)? I’ve been experimenting with structuring this at the scene level—curious what others are seeing.