xAI upgraded Imagine Video 1.5 with image and voice references, prompt-only video generation, and native 1080p output. The update gives users multiple ways to control video creation: describe shots without starting images, guide generation with reference images, or combine character images with voice samples.

Image and voice references launched in the United States for SuperGrok Heavy and SuperGrok Plus subscribers on Grok's Imagine website and iOS app. The company said the tools will roll out to all subscription tiers over the next few days.

Text-to-video and native 1080p generation are already available through Grok Imagine on web, iOS, and Android platforms. The 1080p support works for both text-to-video and image-to-video workflows.

Multi-reference control and voice consistency

The Multi-Reference feature lets creators assign up to seven separate visual anchors to a single generation. One image can preserve a face, another can maintain a product, and a third can define a location. Users can keep characters while changing scenes, retain scenes while swapping characters, or preserve both while altering only the action.

Voice consistency adds another control layer. When a character image and voice reference are supplied together, the model maintains the same face and voice across multiple scenes. This targets a production need: generating multiple shots around recurring characters without rebuilding identity cues for each clip.

Developers can access image references, text-to-video, and native 1080p through the xAI API using the grok-imagine-video-1.5 model. Voice references require a separate request to xAI.

The rollout extends last month's Imagine Video 1.5 release, which xAI positioned as an advance in motion, physics, and audio generation. The latest additions shift focus toward tighter control over identity, scene, and product elements while expanding the model from image-led animation to videos generated directly from text prompts.