Google has announced two updates for its Vids video creation tool: Gemini Omni integration for AI-assisted editing and personal avatars that let users star in videos without filming themselves. These features aim to make video production faster and more accessible.
SynthePulse Insight · AI deep reading
Version 2 · 2 sources
Google Vids introduces the Gemini Omni model and personalized digital avatar features, supporting step-by-step video editing via natural language. The feature is currently available only to Google AI Pro/Ultra subscribers and Workspace business customers, and digital avatars are restricted by age and region.
Google announced a major update to Google Vids on July 16, 2026, centered on the introduction of the multimodal AI model Gemini Omni. According to reports from the Google AI Blog, this feature allows users to generate videos by combining written prompts (text prompt) and uploaded reference images. This capability marks a shift in video creation methods, moving from traditional script-driven approaches to more flexible visual-reference-driven workflows.
Gemini Omni is used not only for initial generation but also introduces 'step-by-step edits' functionality. TechCrunch notes that users can modify already generated videos using natural language instructions, such as changing backgrounds, fixing lighting issues from smartphone footage, or adding special effects. This interaction style allows users to make real-time adjustments during the creative process without having to regenerate from scratch, enhancing workflow flexibility.
Another key update is the personal avatar feature. Users can create a digital stand-in that appears in videos by uploading a selfie and a voice recording. Google emphasizes that these avatars are tied to the user's Google account and are limited to the likeness of the account holder. To prevent misuse, all AI-generated clips are embedded with an invisible SynthID digital watermark to verify content origin.
This update is seen as a signal that Google Vids is transitioning from a single-purpose AI-assisted workplace presentation tool to a full-featured video creation platform. By integrating Vids into Google Workspace, Google intends to strengthen its application in business scenarios such as corporate training and company news updates. However, personalized avatars and conversational editing features also place it in direct competition with startups specializing in AI video generation, such as HeyGen, Synthesia, and D-ID.
The availability of new features is strictly limited. Gemini Omni and digital avatar features are open only to individual users with Google AI Pro and Ultra subscription services, as well as commercial customers of Google Workspace. Additionally, for compliance and security reasons, the personal digital avatar feature is limited to users aged 18 and above and is only available in specific regions. The list of specifically supported regions has not been detailed in public materials.
TechCrunch reported that since the full launch of Veo 3.1 in February 2026, Google Vids has cumulatively generated millions of videos. This data is a media citation, marked as source_claim, and has not been independently verified by Google. Furthermore, there is no detailed information disclosed regarding the robustness of the SynthID watermark after video compression or secondary editing, nor the actual performance of Gemini Omni in complex long-video scenarios, leaving some uncertainties.
This article is based on reports from the Google AI Blog and TechCrunch. The Google AI Blog provides first-hand product feature descriptions, while TechCrunch offers third-party market analysis and partial data citations. All factual statements correspond strictly to the source material, with no external knowledge introduced.
Google Vids enhances the flexibility and personalization of video creation through Gemini Omni and digital avatar features, but its commercial application remains limited by subscription tiers and regional compliance requirements. Users should be aware of the presence of SynthID watermarks and the current access thresholds for these features.
Primary report
Primary source
Also covered by 1 sources