Gemini Omni 1.1 Flash adds scene extension, frame interpolation, and 4K upscaling for generative video

Gemini Omni 1.1 Flash introduces three concrete changes for generative video workflows. Scene extension now allows developers to lengthen clips in 10-second increments, up to a total of 40 seconds, while retaining 10 seconds of prior context, an order-of-magnitude improvement over previous models limited to 1-second lookback. This reduces visual discontinuities when branching narratives or extending sequences, but the 40-second cap may still require manual stitching for longer projects.

The update adds frame interpolation for camera control, letting developers specify start and end frames to generate smooth transitions like dolly zooms or orbital rotations. This shifts creative control from post-processing to the generative stage, but the quality of interpolated movements depends on prompt precision. Overly complex prompts may produce artifacts, requiring iterative refinement or fallback to traditional keyframing.

A two-stage workflow is now possible: 360p previews for rapid iteration, followed by 4K upscaling for final output. This reduces compute costs during prototyping but ties developers to Google’s API for both stages. The 4K upscaling is positioned as studio-quality, yet the material does not specify whether it uses temporal data or frame-by-frame processing, leaving potential limitations for fast-moving scenes unaddressed.

The update targets professional deployment, with integration via Google AI Studio or the Gemini Enterprise Agent Platform. This suggests a push toward enterprise adoption, but the lack of on-premise or hybrid options may limit use cases in latency-sensitive or air-gapped environments. The material also does not clarify whether the model supports custom fine-tuning, which could be a barrier for teams needing domain-specific adjustments.