Amazon has upgraded its AI video model, Nova Reel, to now generate videos up to two minutes long.
First launched in December 2024, Nova Reel marked Amazon’s entry into the generative video space, where it competes with models from companies like OpenAI and Google in an increasingly crowded market.
The latest version, Nova Reel 1.1, offers the ability to generate “multi-shot” videos with a “consistent style” across shots, according to AWS developer advocate Elizabeth Fuentes in a blog post. Users can now provide a prompt up to 4,000 characters long to create videos lasting up to two minutes, made up of six-second shots.
Nova Reel 1.1 also introduces a new mode called “Multishot Manual.” In this mode, users can provide an image along with a prompt to gain more control over the composition of each shot in the video. Fuentes explained that with a 1280 x 720 resolution image and a 512-character prompt, Multishot Manual can produce videos containing up to 20 shots.
Nova Reel is available exclusively through AWS platforms and services, including Bedrock, Amazon’s AI development suite, and users must request access. Like many generative AI systems, there are concerns about whether Nova Reel was developed with ethically sound practices.
Generative video models are trained on large datasets of existing videos to learn patterns and generate new clips. Some companies have faced criticism for training models on copyrighted content without permission from the original creators, which could expose users to potential intellectual property lawsuits when these models generate videos that mimic copyrighted material.
Amazon has not disclosed the specific sources of data used to train Nova Reel or provided a clear method for content creators to opt out of having their videos used in the training datasets. However, the company has stated that it will protect AWS customers accused of copyright violations related to media generated by its models, in line with its indemnification policy.
