TechPrufer
Get started
DeepMind
DeepMind logoDeepMind · Feature trail

Generative Video Models (Veo)

1 post · Jun 2025 – Aug 2026

What this trail covers

Veo continues to evolve with enhanced capabilities for video generation. Veo 3.1 introduces 'Ingredients to Video' with improved expressiveness, character, background, and object consistency, enabling more cohesive narratives. It now supports native vertical outputs for mobile-first content and offers state-of-the-art upscaling to 1080p and 4K resolutions for high-fidelity production. These updates are integrated across consumer and professional platforms, including YouTube Shorts, Gemini app, and now includes personalized video generation, precise motion matching, and the ability to blend live-action and generative footage seamlessly.

All DeepMind trails
  1. Gemini Omni 1.1 Flash lets you build with more controlGemini Omni 1.1 Flash introduces several new generative video capabilities: scene extension (analyzing up to 10 seconds of prior context for improved visual consistency and narrative adherence, with cumulative length up to 40 seconds), specifying first and last frames for smooth transitions and camera movements, faster 360p draft video generation (up to 60% faster and at a third of the cost of 720p), upscaling to 1080p or 4K resolution, and referencing up to three seconds of video in multimodal input for maintaining visual context and character consistency.
  2. Veo 3.1 Ingredients to Video: More consistency, creativity and controlVeo 3.1 introduces 'Ingredients to Video' with improvements in expressiveness, character identity consistency, and background/object consistency. It adds native vertical output (9:16 aspect ratio) for mobile-first content and state-of-the-art upscaling to 1080p and 4K resolutions. These features enhance creative control and production quality for video generation.
  3. Behind “ANCESTRA”: combining Veo with live-action filmmakingThis post details the development of new Veo capabilities to enable greater personalization, precise motion matching, and the ability to blend live-action and generative footage. This includes fine-tuning an Imagen model to match reference image styles for personalized video generation, using a virtual 3D model to record draft shots for motion matching, and developing an 'add object' capability to seamlessly integrate AI-generated elements into live-action footage.