Google Research 推出 AI 视频联合导演：4 个智能体框架实现连贯的分钟级视频生成

Google Research has introduced an AI video co-director for long-form video generation. The suite of 4 agentic frameworks turns short clips into coherent, minutes-long stories. It targets identity drift and cascading errors, the 2 failures that break most multi-shot AI video pipelines today.

Diffusion models render high-fidelity clips in seconds. Stitching those clips into a story is harder. Most agentic pipelines chain modules with independent, handcrafted prompts. That causes semantic drift , where attire or scenery shifts between shots. It also causes cascading failures , where one bad upstream asset corrupts every later shot.

Google team frames this as a credit assignment problem. A broken final video is hard to trace back to the prompt that caused it.

The system sits on top of Gemini and Veo. It is model-agnostic, so the same layer can drive other generators. Outputs inherit SynthID watermarking from the base models.

Co-Director , accepted at COLM 2026, uses a multi-armed bandit (MAB). An Orchestrator Agent picks a configuration across Creative Strategy, Narrative Mode, and Aesthetic Archetype. A Pre-Production Agent builds the storyboard. Keyframe, Video, and Audio sub-agents produce the media. An MLLM Judge then scores the cut and sends a factored reward back to the bandit.

CANVAS , accepted at EMNLP 2026, tracks characters, locations, and object states as the story evolves. It retrieves stored visual anchors when a scene returns. In Google’s museum heist test, AutoStudio lost the thief’s cap and Gemini-3.1-Pro changed the gemstone. CANVAS kept both consistent.

A²RD (Agentic Autoregressive Diffusion) is a training-free architecture. Each segment runs a Retrieve, Synthesize, Refine, Update loop against a multimodal video memory. The agent switches between extrapolation for new story beats and interpolation for returning entities. Google shared a 10-minute film generated this way.

VQQA (Video Quality Question Answering) generates visual questions for each prompt. VLM critiques act as “semantic gradients” that rewrite the text prompt. It needs no access to model internals. A Global Selection step picks the best video across all iterations, not simply the last one.

Google built 3 new benchmarks. GenAD-Bench has 400 ad scenarios across 200 fictional products from 50 brands. HardContinuityBench stresses scene reappearances and prop state changes. LVBench-C has 120 scenarios where key assets vanish for at least 10 segments before returning.

Sources: linked papers, project pages, and GitHub repositories. Verified September 27, 2026.