Amazon has Upgrade Its AI video model, Nova Reel, with the ability to produce video up to two minutes in length.
Amazon’s first broadcast was the Nova Reel Generator Video Space announced in December 2024. It competes with OpenAI, Google and others models that are rapidly becoming a crowd market.
The latest Nova Reel, Nova Reel 4.1, can create “Multi-Shot” video with “serial style” across the shots, explaining AWS developer Advocate Elizabeth Fuentes in a blog post. Users can provide a prompt of 4,000 characters long to create a two-minute video with a six-second shot.
Nova Reel 1.1 also introduces a new mode called “Multisht Manual”. In this mode, the model may refer to an image with a prompt for further control over the composition of a video shot. According to Fuentes, a 1280 × 720-resolution image and 512-macam-character prompt, multi-manual can create 20 shot videos.
Nova Reel is available through AWS platform and services with AII Development Suite only Bedrock, Amazon and customers must request special access. Like most generators AI systems, there are questions about whether the reel was created using the sound system morally.
Video-producing models are trained in a large number of examples of “learning” patterns of these videos to generate new clips. Some companies train models in copyrighted videos without getting permission from owners or manufacturers and when these models “rebuild” the copyrighted steel, it publishes the users’ IP suit.
Amazon did not disclose the data source of training in the Rel, or it did not provide an obvious way to feed the models probably for the manufacturers to choose their videos. The company, however, says that according to the compensation policy, any AWS accused of copyright violation of the media produced by its models will protect customers.