H3 Max Prompt Lab

H3 Max Image to Video Prompt Tips

When you start from an image, the prompt should explain motion, not re-describe the whole picture.

Prompt pattern

  1. Say what must stay locked from the image (face, product label, palette).
  2. Describe the motion that happens next (camera + subject).
  3. Add lighting continuity notes if the still is day/night specific.
  4. Keep duration short while testing (5–8s).

Example brief: “Keep the product label and red packaging exact. Slow orbit right as condensation beads slide down the can. Soft fridge-bay lighting, subtle hiss sound, 5 seconds, 768P.”

← Back to the H3 Max text to video prompt generator

Planning ceremony beats? See AI wedding video prompts (first look, vows, first dance, and more).

Need a Seedance-class CCTV orbit, bullet-time freeze, or product drop? Copy a brief from AI bullet-time video prompts.

Migrating a Sora cameo / remix-ID habit to stills + camera events? See Sora → Seedance / Kling / Veo prompt remap and the Seedance vs Kling vs Veo cheat sheet. MiniMax path: H3 as a Sora alternative. Wan shot timestamps: Wan 3.0 prompt guide. Export / inventory before the 24 Sep 2026 API hard stop: Sora API sunset checklist. Billing units (Veo per-second, Kling credits, Seedance tokens): Sora API cost vs Veo / Kling / Seedance.