Imagine Video 1.5 with reference support

xAI has added the ability to generate videos using image and voice references, as well as text alone, to its video generation model Imagine Video 1.5. Both text-to-video and image-to-video support native 1080p generation.

By inputting a character image together with a voice reference, you can maintain the same face and voice across multiple scenes. Using up to 7 image references, you can also fix characters, products, or locations in place while changing the scene or action.

Text-to-video and native 1080p are available on grok.com/imagine, iOS, and Android, and in the US, the image and voice reference features are launching first for SuperGrok Heavy and Plus users. Image references, text-to-video, and 1080p generation are also available through the grok-imagine-video-1.5 model in the xAI API, and voice references are provided via a request-based method.