H3 Max TikTok Video Generator: Vertical Prompts That Work

A vertical clip has less horizontal room and less time to explain itself. The first frame should reveal the subject; the first motion should establish what is changing; the sound should reinforce that change without waiting for a long introduction.

H3 Max text-to-video includes a 9:16 option, while image-to-video follows the supplied image. Prepare portrait source images when the destination is a vertical feed.

Compose for the phone, not a cropped landscape

Put the important action near the middle of the frame and leave breathing room near interface overlays. Avoid a wide scene with three equally important subjects: it may look balanced on desktop and unreadable on a phone.

A useful portrait composition often has:

• one subject at a readable scale;
• depth from foreground and background movement;
• vertical motion such as falling, rising, pouring, or revealing;
• clean space where real captions can be added later;
• no critical detail pressed against the lower or right edge.

A vertical prompt with an immediate change

Vertical 9:16 close shot of a tiny paper observatory on a windowsill before dawn. Its folded roof opens upward in the first second and a narrow beam of warm light projects a moving star map across the dark ceiling. Slow tilt following the beam, handmade paper fibers visible, quiet blue room beyond. Sound: paper hinges unfolding, one low mechanical hum, distant morning birds. One continuous shot, no titles or subtitles.

The prompt does not say “make it viral.” It describes the event that might make someone pause.

Three prompt shapes for short-form feeds

1. The impossible material

Start with a familiar object behaving in an unfamiliar but physically readable way: stone folds, glass breathes, paper emits light. Keep the camera simple so the event remains the hook.

2. The satisfying process

Show one process completing cleanly: a mechanism aligns, ingredients settle, fabric wraps an object, or a miniature scene assembles. Let synchronized foley carry the rhythm.

3. The spoken observation

Frame one person clearly and give them one short line. Describe voice distance, room tone, and the reaction after the line. A five-second clip cannot comfortably hold a paragraph.

Browse the H3 Max prompt guide for longer examples and the audio guide for dialogue and sound direction.

Text belongs in the edit

Generated titles may drift, change, or compete with platform captions. Ask for a clean visual area, then add the real hook, subtitle, offer, and disclosure in an editor. This also lets you reuse the same clip across placements with different copy.

A sensible generation sequence

1. Generate five seconds at 480P.
2. Watch once without sound: does the action read?
3. Listen without watching: is the sound clean and relevant?
4. Check the frame at phone size.
5. Change one weak element.
6. Move the selected version to 768P.
7. Add captions and brand elements outside the generated footage.

Use eight to ten seconds only when the idea needs a second beat. Duration should earn its place; a longer clip is not automatically a better feed clip.

Before posting

• Verify music, likeness, product, and source-image rights.
• Check any promotional claims and required disclosures.
• Add accurate captions if the clip contains speech.
• Preview the final export inside the actual placement interface.
• Keep the original generation separate from the edited master.

This page describes a production workflow, not a promise of reach. Distribution depends on the finished idea, account, audience, timing, and the platform’s current systems.