Google’s Veo 3.1 brings vertical video, higher-fidelity upscaling and better character consistency

Google has rolled out Veo 3.1 upgrades to its ‘Ingredients to Video’ capability, introducing native vertical outputs and sharper upscaling options while promising more consistent characters and scenes across clips.

The update targets creators who design videos from reference images and are increasingly optimising for mobile-first, short-form formats, according to the company.

More consistency and creative control

Veo 3.1 aims to make image-to-video outputs feel livelier and more coherent. Google says identity consistency for recurring characters has improved, background and object details can be kept steady across shots, and disparate elements such as textures, characters and stylised environments can be blended into a single clip.

These changes are designed to help both casual storytellers and professional teams produce narrative sequences with fewer visual jumps.

How will vertical video support change creator workflows?

The update adds native 9:16 portrait generation so shorts can be produced for platforms like YouTube without post-cropping, reducing quality loss and saving edit time. Analysts said this aligns with the shift to phone-centred viewing, where vertical framing is often the default for discovery and engagement. Google positions the feature as mobile-optimised output for short, snackable storytelling.

Where you can try it

The company is bringing Ingredients to Video directly into YouTube Shorts and the YouTube Create app for the first time. Enhanced Veo 3.1 features, including portrait mode, are available in the Gemini app starting now.

For professional workflows, the rollout covers Flow, the Gemini API, Vertex AI and Google Vids. Google notes that upscaling to 1080p and 4K is supported on Flow, the Gemini API and Vertex AI.

• Consumers and creators: Gemini app, YouTube Shorts and the YouTube Create app
• Enterprises and studios: Flow, Gemini API, Vertex AI and Google Vids
• Higher-fidelity delivery: 1080p and 4K upscaling in Flow, the API and Vertex AI

Watermarking and verification

All videos made with Google’s tools carry SynthID, the company’s imperceptible digital watermark. In December 2025, Google expanded the Gemini app to verify videos by scanning for SynthID in audio and visuals, so users can ask if a clip was generated with Google AI.

The feature supports uploads up to 100 MB and 90 seconds, and is available wherever the Gemini app is supported.

What it means for the creator economy

For India’s large base of smartphone-first creators and early-stage brands, vertical-by-default generation lowers friction in producing Shorts-style explainer videos, product demos and quick social promos. Improved character and background consistency can reduce reshoots and manual masking in post, while higher-fidelity upscaling helps when repurposing clips for larger displays.

Founders experimenting with AI-first workflows could stitch short clips into serial narratives, test multiple visual treatments, and then move successful concepts into longer edits using conventional tools.

Inside Google DeepMind, led by co-founder and CEO Demis Hassabis, research-to-product handoffs such as SynthID watermarking and verification, alongside model upgrades like Veo 3.1, are increasingly framed as responsible innovation meant to balance capability with provenance.

Creators, however, are still expected still apply standard content checks, brand-safety reviews and consent protocols when using synthetic footage in campaigns.