Amazon's Nova AI Can Now Generate Minutes-Long, Multi-Shot Videos

Amazon Steps Up in the Generative Video Race

Amazon Web Services (AWS) has introduced a major update to its generative AI tool, Nova, now capable of producing multi-shot, two-minute-long videos based on text prompts. The update to Nova 1.1 significantly enhances its capabilities, putting Amazon in direct competition with OpenAI’s Sora and Google’s Lumiere in the rapidly advancing world of AI video generation.

From Snippets to Full Scenes

First launched in December 2024 as Nova Reel, the model was originally limited to shorter, simpler clips. With the new version, users can now generate videos in 6-second segments that are automatically stitched together, allowing for up to 20 scenes per clip, all driven by a single prompt of up to 4,000 characters.

These stitched scenes create a cohesive visual story, opening the door for creators, developers, marketers, and educators to generate complete videos with minimal technical effort.

More Creative Control With "Multishot Manual"

In addition to full-length generation, Amazon also unveiled a new Multishot Manual mode, giving users unprecedented control over each scene. In this mode, users can input separate prompts of up to 512 characters per shot and include reference images to better guide the model’s output.

This makes Nova especially useful for projects requiring narrative continuity or specific visual aesthetics — a feature not fully offered by many of its competitors.

Built for Developers and the Cloud

Nova 1.1 is now available via Amazon Bedrock, a platform within AWS that provides access to foundation models and generative AI services. Access to Nova remains open to developers by request, and according to AWS, access is automatically approved for those who apply.

Each generated video is rendered in HD resolution (1280x720), optimized for both previewing and real-world use in digital content.

Ethical Questions Still Loom

While the tech world applauds the capabilities of Nova 1.1, concerns persist about how Amazon trained the model. Like others in the generative AI space, Nova likely relies on vast datasets — potentially including copyrighted material — although Amazon has not publicly disclosed the exact sources.

To help ease these concerns, Amazon reinforced its Customer Agreement, which includes a commitment to defend users against legal claims if content generated with Nova triggers copyright disputes. Still, many industry experts are calling for greater transparency.

A Sign of What’s to Come

With this release, Amazon enters a growing market of AI tools reshaping creative industries. OpenAI’s Sora and Google’s Lumiere are already making waves with lifelike video generation, and Amazon is positioning Nova as a more customizable, developer-friendly, and cloud-integrated alternative.

The advancements in Nova point to a future where AI tools won’t just assist in video editing — they’ll generate entire narratives from scratch, changing how media is made and consumed.





News Source: TechCrunch