Amazon has rolled out a major upgrade to its generative video model, Nova Reel, expanding its capabilities to produce videos up to two minutes in length and offering users greater creative control.
Launched in December 2024, Nova Reel marked Amazon’s entry into the competitive generative video space, joining offerings from OpenAI, Google, and others. The latest version, Nova Reel 1.1, was announced in a blog post by AWS developer advocate Elizabeth Fuentes, who described the update as a leap forward in video generation technology.
Nova Reel 1.1 supports “multi-shot” video creation, allowing for a consistent visual style across scenes. Users can now input text prompts up to 4,000 characters long to produce videos composed of six-second segments, stitched together to form a cohesive two-minute clip.
The new release also introduces “Multishot Manual,” a mode that gives users enhanced control over video composition. In this mode, users can submit an image alongside a prompt (up to 512 characters) to generate videos made up of as many as 20 individual shots. The model supports reference images with resolutions up to 1280 x 720.
Nova Reel remains exclusive to Amazon Web Services platforms, including the Bedrock AI development suite. Access to the tool is granted by request.
As with other generative AI tools, questions remain about the ethical sourcing of training data. Video generation models are typically trained on large datasets of video content, and some AI firms have come under scrutiny for using copyrighted material without permission. Amazon has not disclosed the specific datasets used to train Nova Reel, nor has it provided a mechanism for content creators to opt out of having their work included in future training sets.
However, Amazon says it will protect AWS customers from legal liability should the AI-generated content produced using Nova Reel result in intellectual property disputes. This pledge aligns with the company’s broader indemnification policy for its AI services.
