xAI officially announces the launch of Imagine Video 1.5, its best video model to date, now adding text, image, and audio reference capabilities, and capable of generating video up to 1080p.
New text-to-video functionality and native 1080p support
Imagine Video 1.5 now supports text-to-video functionality, allowing users to describe scenes without a starting image. This technology combines image generation with image-to-video capabilities, and native 1080p is now available for both text-to-video and image-to-video generation.
“The Imagine Video 1.5 we’re releasing is our best video model to date, with better motion, better physics, and better audio.”
xAI
Launch of audio consistency feature
Users can now provide character images and audio references, ensuring that character faces and voices remain consistent across every scene. This feature significantly improves video coherence.
“Pass in a character image and an audio reference, and both stay consistent in every scene.”
xAI
Multi-reference functionality applications
Each reference image can lock a specific element, such as a face, product, or location. Users can swap scenes while keeping the character unchanged, or swap characters while keeping the scene unchanged, with up to seven references usable simultaneously for generation.
“Each reference image locks an element, allowing users to flexibly adjust characters or scenes.”
xAI
Feature updates in the API
In xAI’s API, image references, text-to-video, and native 1080p functionality are now live, along with support for the best video model, grok-imagine-video-1.5. Audio reference capability is available on demand.
Source: xAI official announcement