A tall vertical frame glowing in a dark studio with motion trails around it

Vertical Video

Grok Imagine for TikTok, Reels, and Shorts: A 9:16 Workflow

Vertical video is not a cropped landscape clip. The viewer holds the frame at arm’s length, the first second decides whether the rest exists, and every important detail has to survive a thumb, a caption, and a user interface. This guide turns Grok Imagine into a practical TikTok, Reels, and Shorts workflow: one hook, three shots, a safe 9:16 composition, and a clean handoff to the editor where captions and sound belong.

Updated 2026-10-19Jump to FAQ

Why vertical changes the story

A landscape frame invites the eye to wander left and right. A vertical frame stacks information: foreground at the bottom, subject in the middle, destination at the top. That is why a simple reveal, fall, turn, or rise reads so well on a phone and so poorly when cropped from a wide shot.

Plan for the final shape from the first generation. Use 9:16, keep the subject inside the central column, and leave the top and bottom for platform overlays. A clip that already feels vertical needs fewer fixes after download.

The one-second rule: The opening frame must show a subject, a change, or a question. If the first second is only a camera move through empty space, the viewer has no reason to wait for the payoff.

Three hooks that work in one second

A hook is not a slogan. It is a visible reason to keep watching. These three structures are easy to generate and easy to cut around.

The interruptionStart with the result already in motion: a glass shattering, a door bursting open, a product spinning into frame. Open on the most legible half-second, then rewind or explain.
The impossible close-upBegin very close on a texture or detail, then pull back to reveal the whole object. The viewer stays for the reveal because the first frame withheld context.
The direct addressA character looks at camera and begins an action, with no establishing shot. This works when the face and gesture are clear at phone size and the background is quiet.

The three-shot structure

Most short-form clips need three jobs, not thirty shots: catch attention, prove the idea, and land a payoff. Generate each job separately so a weak beat does not contaminate the whole sequence.

Hook: 0 to 1 secondUse one strong image or action. Keep camera movement minimal and put the subject in the middle of the vertical frame.
Proof: 1 to 4 secondsShow the change, the product, the process, or the second angle that makes the hook believable. One idea per shot.
Payoff: 4 to 7 secondsLand the result, expression, or final angle. Leave a clean half-second of usable tail so the edit does not end on an unfinished motion.

Compose for the interface, not just the video

Every short-form platform covers part of the frame. Captions sit low, controls sit right, and usernames or descriptions can occupy the bottom-left. The safe answer is not a precise pixel map that changes with every release; it is a conservative composition rule.

Place the face and the product in the central vertical third, not against the right edge.
Keep the lower fifth free of essential detail; captions and interface elements will compete there.
Avoid tiny text inside the generated image; it will be unreadable after compression and overlays.
Shoot the subject slightly smaller than feels necessary in a landscape mindset; vertical video has less horizontal room.
Check the final clip on a phone before export, not only in the desktop preview.

A tall luminous frame with generous dark margins around it — Keep faces and product details inside the central column; reserve the edges for captions and interface overlays.

Sound and captions belong in the edit

Generation can suggest a sound bed or a voice direction, but it is not a captioning or publishing tool. Build the picture here and finish the audio grammar in an editor where timing, loudness, and text placement are predictable.

TaskInside this studioIn the editor

Shot and motionGenerate clips at 9:16Trim, order, and pace

Voice or ambienceUse audio references where supportedMix levels and replace weak audio

CaptionsNot a text timelineAdd, style, and time the words

Hook and retentionGenerate the visual beatPlace the cut and test the opening

ExportDownload the clean clipEncode the master for each platform

Five short-form templates

Change the subject and setting, keep the structure. Each pattern is sized for a single generation or a small cluster of clips.

TemplateOpening imagePayoff

Product turnMacro detail of a matte surfaceProduct rotates into a clean hero angle

Before and afterMessy or neutral first stateSame frame transformed by one visible change

One-take walkCharacter enters a corridorCamera follows to a reveal around a corner

Impossible scaleTiny hand holds an objectCrane up to reveal a giant environment

Loop teaseMotion already in progressFrame returns near its starting position for a seamless repeat

Export and quality checklist

The platform will recompress what you upload, so give it a clean master and avoid creating avoidable softness before delivery.

Export the tallest practical version of the final edit and keep the original generated file as a backup.
Use the same frame rate across shots; mixed rates can stutter after platform processing.
Add captions after generation and keep them inside the safe central column.
Watch the clip once with sound off and once with sound on; the hook must work both ways.
Name files by campaign and shot number so a replacement does not overwrite the wrong asset.

A vertical clip is a composition problem first: Generate for 9:16, cut for the first second, and only then worry about export settings. A perfect bitrate cannot rescue a frame that was composed for landscape.

Three tall glowing frames stacked in depth in a dark room — Three vertical beats are usually enough: hook, proof, payoff.

Frequently asked questions

Does Grok Imagine support 9:16 video?+

Yes. The studio lists 9:16 among its aspect-ratio options. Choose it before generating, because re-framing a landscape render later usually loses the composition that made the shot work.

How long can a TikTok video from Grok Imagine be?+

A single generation is 4 to 15 seconds. Build longer short-form videos by generating several beats and joining them in an editor; use extend only when one action must remain continuous.

Can I add captions inside the studio?+

No. The studio does not provide a caption track or a full text timeline. Add captions after download in an editor, where you can control font size, safe areas, and timing.

Why does the clip look softer after uploading to TikTok?+

Short-form platforms recompress uploads. Start with a clean, high-quality master, avoid excessive denoising or sharpening, and keep the subject large enough that compression does not destroy the detail.

Should I make one 15-second clip or three five-second clips?+

For most short-form storytelling, three short beats give you more control over the hook, proof, and payoff. Use one continuous clip when the action itself is the reason to watch.

Can I use the same footage for Reels, Shorts, and TikTok?+

Yes, if the 9:16 frame is composed conservatively. Keep essential detail away from the edges and re-check captions and interface overlays in each platform because the safe areas are not identical.

Related guides

Build for the First Second

Generate a clear hook, stack three purposeful beats, and finish captions and sound in the edit.

See Plans