Amazon has an upgrade for its AI video model, Nova Reel. It can now generate videos up to two minutes in length. Launched in December, it was the company’s first foray into the generative video space, which has players like OpenAI and Google.

Nova Reel 1.1 allows users to create “multi-shot” videos with consistent styling across segments. It now supports prompts of up to 4,000 characters. There is also a new mode called Multishot Manual, which can reference an image along with a prompt to offer more control over a video shot’s composition.

According to TechCrunch, Nova Reel is available through AWS platforms and services, including Bedrock, Amazon’s AI dev suite, and customers must request access.

Video-generating models are trained on a wide range of examples of videos to “learn” patterns to generate new clips. Some companies train models on copyrighted videos without obtaining permission from the owners, exposing users of the models to IP lawsuits. Amazon has not mentioned the source of Nova Reel’s training data. But TechCrunch reports that Amazon has said it’ll protect any AWS customers accused of violating copyright with media generated by its models, in keeping with its indemnification policy.