Gemini Omni Prompts
Home / All Prompts / cinematic

Cinematic Continuous Shot — Designer's Rainy Window

Adapted from PixVerse's hands-on Gemini Omni test — uses 'one continuous shot' + duration + aspect ratio in the opening line.

cinematic 16:9 10s PixVerse-tested structure #cinematic#continuous-shot#push-in#mood#rainy-window

Prompt

Create a 10-second 16:9 cinematic video in one continuous shot. A young product
designer sits at a small desk beside a rainy window in a softly-lit apartment.
She sketches in a notebook while occasionally glancing at the rain. The camera
slowly pushes in from a wide shot to a close-up of her hands and the notebook.
Warm tungsten lamp light contrasts with cool blue rain outside. Subtle ambient
sound, slight handheld micro-jitter for realism.

Why this prompt

Structure adapted from PixVerse’s hands-on Gemini Omni review, which actually tested this prompt on Omni Flash. Their finding: “strong on mood and texture; weaker on prompt completion” — meaning Omni nails the atmosphere but may de-emphasize secondary objects you mention.

Source tier: 🟡 Media hands-on (medium-high confidence — actual output exists at PixVerse)

The opening-line formula

The first line of your prompt sets the structural frame. Per PixVerse’s testing, this pattern works reliably:

Create a [duration]-second [aspect-ratio] [genre] video in one continuous shot.

Why “one continuous shot”: Omni interprets this as “no cuts, no edits” — which prevents the model from injecting unwanted transitions and keeps your scene coherent.

Camera vocabulary that Omni understands

From DeepMind’s prompt guide, Gemini Omni explicitly recognizes these terms:

  • Camera motion: push in / punch in / dolly zoom / pull-back / orbit / sweep
  • Shot type: oner / continuous shot / locked off / fixed / over-the-shoulder
  • Style: natural smartphone zoom / film camera / webcam style / handheld

Use these instead of vague phrases (“epic camera move”) — Omni parses them as specific instructions.

How to tweak

  • Setting swap: rainy window → snowy cabin / sunlit attic / submarine porthole
  • Subject swap: designer → musician tuning instrument / chef tasting / artist mixing paint
  • Mood pivot via lighting: “warm tungsten + cool blue rain” → “candlelit + storm flashes” / “neon cyberpunk + holograms” / “morning fog + golden hour”
  • Camera intent: “slow push-in to close-up” → “static wide” / “slow pull-back reveal” / “orbit around subject”

Conversational follow-up (Omni’s superpower)

After the base shot, you can iterate:

Turn 2: Change the camera to be over the designer's shoulder, framing the notebook
from above. Keep the lighting and her movement consistent.

This pattern (lock one variable, change one variable) is from Medium’s Gemini Omni Prompt Playbook.

Common failure modes

  • Over-prescription: DeepMind explicitly says “prompts don’t have to be as prescriptive” as Veo — avoid stacking 6+ adjectives. Natural language works better.
  • Word count > 50: per seaart’s testing, prompts over ~50 words dilute focus
  • Secondary objects ignored: per PixVerse, if you mention a “drone hologram” as background detail, Omni may skip it — don’t rely on minor objects landing

Sources

Related prompts