Gemini Omni adds conversational video generation, editing

This is more interesting as an editing workflow breakthrough than as another text-to-video demo. If Google’s consistency claims hold up outside polished demos, Gemini Omni could move video AI from prompt lottery to iterative production tool.

• –Conversational edits matter because real creative work is revision-heavy, not one-shot generation.
• –Reference-based creation plus stronger scene consistency should reduce drift across characters, shots, and style.
• –Supporting text, image, audio, and video inputs makes it a broader multimodal creation layer, not just a generator.
• –Bundling across Gemini, Flow, and YouTube gives Google distribution leverage that standalone video startups do not have.
• –The open question is temporal coherence across multiple edits; that is where most video models still break down.