Amazon has updated her video model he, Nova Reel, with the ability to generate video up to two minutes in length.
Nova Reel, announced in December 2024, was Amazon’s first form in the video generating space. She competes with models from Openai, Google, and others in what is quickly becoming a crowded market.
The latest Nova Reel, Nova Reel 1.1, can generate video “with” many strokes “” sustainable style “across shots, explained AWS developer Elizabeth Fuentes in a blog post. Users can provide a tall up to 4,000 characters long to generate up to a two-minute video consisting of six-second shots.
Nova Reel 1.1 also introduces a new way called “Multishot Manual”. This way, the model can refer an image along with a quick to offer more control over the composition of a video. According to Fuentes, given a resolution of 1280 × 720 and 512-maximum-rapid character, multiple manuals can generate videos containing up to 20 shots.
NOVA Reel is only available through AWS platforms and services, including Bedrock, Amazon Development Suite, and customers should require special access. As with his most generating systems, there are questions whether Reel has been developed using sound ethical methods.
Video generation models are trained in a large number of video examples to “learn” models in these videos to generate new clips. Some companies are training models into copy -protected copyright videos without receiving permission from owners or creators, and when these models “adjust” copyright -protected photographs, it exposes users of models in IP lawsuits.
Amazon has not revealed Reel’s training data source, nor has it provided a clear way to creators who can feed the models data with their videos to give up. However, the company has said it will protect any AWS clients accused of violating copyright with media created by its models, in accordance with its indemnity policy.