Grok Imagine Cinematic Language Visual Notes: Image-to-Video Camera Movement Demonstration

Starting from a single static storyboard frame, this uses grok-imagine/image-to-video to demonstrate the visual differences among the four basic camera movements: push in, pull out, pan, and truck. The reference image is deliberately made as a top-down floor-plan annotation of the same scene: it includes a mid-ground figure, foreground occluders, and a deep corridor, making it easy to observe the perspective changes caused by camera displacement. When submitting, pass a single reference image via image_urls, use prompt with @image1 to point to that image and specify the movement direction clearly, set resolution to 720p, duration to 6 seconds, and mode to normal, and the output aspect ratio follows the input image. It is suitable for storyboard artists and short-video directors to quickly check camera-movement intent before formal generation, and also for developers to understand how to write motion control for this model under single-image input. This model family is positioned as integrated image-to-video generation, and the duration of a single segment is limited, so this work only uses a single shot and a single action, and does not stack multiple camera movements in one clip.