Prompt Grok Imagine for image-to-video

Last updated: October 12, 2026

Grok Imagine Video animates an approved still into a short clip with sound, and holds the character and the art style through fast action. Pick it for ability casts, attacks and impacts, where several beats have to land in a few seconds.

Let the agent write it

You don't have to write any of this yourself. The Layer agent knows how to prompt each model it runs, from the same guidance this page is built on. Attach the still in chat, name the model and say what should happen, for example: Animate this with Grok Imagine Video 1.5: the mage leaps and slams his staff into the ground in a burst of fire, locked camera, 3 seconds, with the sound of the blast. The agent writes the prompt, shows you its plan, and runs it when you approve. Use the rest of this guide when you want to write or adjust the prompt yourself, and see Write a brief for the agent for how to brief it.

When to use it

Use Grok Imagine Video when the still is signed off and the job is motion: attacks, spell casts, power-ups, idles and short trailer beats. It keeps the camera where you put it and the character on-model, even through a leap and an explosion, and it generates sound in the same pass.

Keep titles and logos out of the clip, and check hands before a clip ships; both are weak spots. For timed multi-shot clips, see Prompt Seedance for image-to-video. To generate a clip with no image at all, see Create a video.

Versions on Layer

Grok Imagine Video 1.5

The version this guide's examples were made with. It is built for image-to-video: it needs a First frame, renders at 480p or 720p, and runs 1 to 15 seconds. Use 480p to find the motion, then render the keeper at 720p.

Grok Imagine Video

The original model, at 720p only and 1 to 15 seconds. It also works without an image (text to video) and can edit an existing clip from a Source video. On the same frame and prompt it moved much like 1.5: same locked camera, same character, a slightly different take on the fire.

How to write the prompt

Camera, then the action in the order it happens, then the style, then a Sound: line:

Locked, static camera. The mage explodes into action, leaps high into the air, spins and slams the staff into the ground, unleashing a violent fiery explosion all around him. Stylised painterly game art, colours unchanged. Sound: a roaring blast of fire and cracking stone.

The first frame already fixes who is in the shot, the composition, the light and the look, so the words go on what changes. Grok renders a clip from the first frame forward, so write the beats in the order they should happen: something mentioned last can arrive late or not at all. That prompt is the one in the clip above.

What works

Several beats, in order

Write the action as a sequence with strong verbs: "leaps high into the air, spins and slams the staff into the ground, unleashing a violent fiery explosion". In our test, a 3-second clip landed every beat in order (crouch, leap, slam, a ring of fire and flying rock) with the camera locked and the mage on-model through the explosion. Intensity words such as "with tremendous force" and "explodes upward" read as how hard things move.

A still camera by default

The camera holds still unless you ask it to move. We ran the same prompt with and without "Locked, static camera" on the same frame, and both clips kept the camera locked. When you want a move, name one: slow push-in, slow pan out, tracking shot alongside the mage, handheld. Describe it concretely ("the camera arcs a quarter of the way around him") rather than "cinematic camera work".

Motion, not description

Spend the prompt on what moves and how. About three sentences of motion is enough; detail spread across many objects dilutes it. End with the medium ("stylised painterly game art, colours unchanged") so stylised art stays stylised.

A Sound line

Every clip comes with sound, so direct it. Close the prompt with Sound: and name materials and the space: "boots crunch on gravel to the left", "wind gusting across the stone ledge", "muffled through glass". Say whether you want music. Put any spoken line in double quotes and keep it short.

Say it positively

Write what you want rather than what you don't: "locked, static camera" says in three words what "the camera does not move: no zoom, no pan, no push-in" says in nine. In our test the negated version was followed too, but the positive one is shorter and leaves less to misread.

A start image that can move

Pick a still with one clear subject, readable hands and limbs, separation from the background and room in the direction the action goes. A frame that could be the first frame of a shot animates better than a crowded composition.

What to avoid

Text, hands and exact likeness

In-frame lettering, hands and a character's exact face can wobble. Paint titles into the still or add them afterwards in the video editor, and review hands before the clip goes into a build.

Expecting a different shape from the image

The clip follows the shape of the first frame. Crop the still to the shape you need before you animate it.

Settings that matter

First frame is the still the clip starts on. Duration runs from 1 to 15 seconds; find the motion at a few seconds, then render the length you need. Aspect ratio sets the resolution; on Grok Imagine Video 1.5, Auto follows the first frame at 480p or 720p. Above the prompt, Camera, Staging, Look and Performance take reference sets for each.

Prompts to start from

Ability cast

Locked, static camera. The mage explodes into action, leaps high into the air, spins and slams the staff into the ground, unleashing a violent fiery explosion all around him. Stylised painterly game art, colours unchanged. Sound: a roaring blast of fire and cracking stone.

Power-up idle

Locked, static camera. The mage slowly raises his staff and the ember at its tip flares brighter, sparks spiralling upward, his cloak stirring in the wind. Stylised painterly game art, colours unchanged. Sound: fire crackling at the staff tip, wind gusting across the stone ledge.

Use it in Layer

Add the approved still to a session's board.
Click View on the image, then Turn into video. Create video opens in the side panel with the image as its First frame.
Click the model name at the top of the panel and choose Grok Imagine Video 1.5. The first frame stays in place.
In the prompt box, write the camera, the action beat by beat in the order it happens, the style and a Sound: line.
Set Duration to a few seconds while you find the motion.
Click Create video. The cost shows on the button before you click.
When the clip lands on the board, click View to open it in the viewer and play it. Then change one thing in the prompt at a time and run it again.

Learn more

Video generation (xAI): the video models, durations and resolutions, that image-to-video follows the image's shape, and that sound is generated by default.
How to use Grok Imagine (fal): subject and action as the two anchors, concrete camera moves, and how to direct the sound and dialogue.
Grok Imagine Video 1.5 image to video (fal): the image-to-video inputs and their defaults.
Grok Imagine prompt guide (thoxakihiko, GitHub): field-tested patterns checked against xAI's API docs, a camera move dictionary, the Sound block, and the anti-patterns.