Grok Imagine (Video)

Grok Imagine Video takes a still photo and animates it into a 30-second clip, adding realistic motion, object interactions, and camera movement while keeping the scene grounded in the original image. It's a good fit whenever you already have the exact frame you want to start from and need the model to bring it to life rather than generate a new scene from scratch.

Its longer 30-second clip length relative to most other image-to-video models on AutorunX makes it useful for shots that need to sustain motion or an action over a longer stretch than a typical 8-15 second clip allows.

Grok Imagine Video is xAI's image-to-video model: it takes a single still image as its starting point and generates motion, camera movement, and object interaction from it, rather than accepting a text-only prompt to build a scene from nothing. There is no text-to-video mode on this model — every generation begins from a supplied image.

The model is built to prioritize consistent, stable motion and controlled camera behavior over dramatic but unpredictable movement, which makes it comparatively conservative in how much a shot changes from its source frame. On AutorunX, Grok Imagine Video produces 30-second clips at 480p-720p resolution.

Because it's image-anchored rather than prompt-anchored, the composition, subject, and setting of the output are locked in by the input image; the prompt text mainly steers what the subject does and how the camera moves within that fixed scene rather than what's in it.