pro tip: google gemini is GOATED at video understanding you can upload video and it understands motion, progression, dialogue, music, etc astra and fable can’t do that. they’ll take screenshots and transcribe voices and try to understand what’s going on. but it’s not the same since astra and fable 5.1 were clearly trained on blender, they can do some moves that they know look good. but outside of their repertoire of party tricks they can’t really analyze the quality of a video they output. they can only analyze screenshots hook up the gemini api into your agent and tell it to use it for video analysis. you won’t regret it