Seedance 2.0 Vertical Video Prompts for Social (9:16 That Actually Composes Well)

Vertical video fails when the model composes for a wide frame and your subject ends up tiny. Fix it by prompting composition, not just content.

The vertical composition rules

• Keep one hero subject centered.
• Keep the horizon low or high (avoid dead center).
• Put secondary details in the top third or bottom third, not both.

Add a "frame" line to your prompt

Paste this line into any prompt:

Framing: vertical 9:16 composition, hero subject centered and large in frame, head-to-toe or chest-up, minimal off-screen cropping.

Template: talking head / portrait (best reliability)

Framing: vertical 9:16, chest-up portrait, eyes sharp, centered.
Camera: slow dolly-in, 85mm look, shallow depth of field.
Lighting: soft key, gentle rim, controlled highlights.
Background: clean bokeh, no distracting motion.

Template: full-body fashion (keep feet in frame)

Framing: vertical 9:16, full-body shot, feet visible, subject centered.
Camera: slow orbit, smooth gimbal, constant speed.
Lighting: editorial studio look, crisp silhouette, glossy highlights.

Template: action (readable in a narrow frame)

Framing: vertical 9:16, wide-enough to keep full action readable, subject stays centered.
Camera: tracking shot that follows the subject, no whip pans.
Action: one clear beat, minimal background clutter.

The social "hook" trick

Make the first second visually clear:

• big subject
• strong contrast
• one obvious motion

Then build from there.