The Complete AI Video Prompt Guide

Every effective AI video prompt follows the same structure:

Subject + Action + Setting + Camera + Style

Bad:  "A cat in a garden"
Good: "A fluffy orange cat walks slowly across a stone pathway,
       through a sunlit Japanese garden with cherry blossoms.
       Camera pans right following the cat. Cinematic, peaceful."

Subject: Be Specific, Not Exhaustive

Pick 3-5 key visual features that matter most.

| Weak | Strong | |------|--------| | A man | A middle-aged detective in a worn grey coat | | A car | A sleek red sports car with chrome details | | A building | A crumbling gothic cathedral covered in ivy |

Action: One Main Movement

Keep it focused. AI works best with one clear action per video.

| Too vague | Too complex | Just right | |-----------|-------------|------------| | A woman moves | A woman jumps, spins 3x, backflips, waves | A woman in a red dress twirls once, her dress flowing, then smiles at camera |

Setting: Time + Weather + Details

Don't just say where — describe what it looks and feels like.

Basic:  "A forest"
Better: "A misty forest at dawn with tall pine trees"
Best:   "A misty forest at dawn, golden sunlight breaking through branches, moss-covered ground"

Include:

• Time of day — dawn, noon, golden hour, dusk, night
• Weather/atmosphere — foggy, rainy, windy, clear
• Specific details — what's in the background
• Lighting quality — harsh, soft, dramatic, natural

Camera Movements: Complete Reference

Basic Movements

| Movement | Prompt Command | Effect | |----------|---------------|--------| | Push In | Camera slowly pushes in toward subject | Emotional intensity | | Pull Out | Camera pulls back to reveal environment | Context reveal | | Pan Left/Right | Camera pans slowly to the right | Horizontal scanning | | Tilt Up/Down | Camera tilts up from feet to face | Vertical scanning | | Truck Left/Right | Camera trucks left with parallax | Lateral with depth | | Tracking Shot | Camera tracks alongside subject | Following movement | | Zoom In/Out | Slow zoom in on subject's face | Optical magnification | | Orbit | Camera orbits 360° around subject | Circular reveal | | Handheld | Subtle handheld shake | Realism, tension |

Cinematic / Special Movements

| Movement | Prompt Command | Effect | |----------|---------------|--------| | Crash Zoom | Fast zoom in on character's expression | Shock, surprise | | Dolly Zoom | Dolly zoom effect, background distorts | Unease, disorientation | | Rack Focus | Rack focus from foreground to background | Attention shift | | FPV Drone | FPV drone flying through narrow gaps | Adrenaline, speed | | Aerial Drone | High altitude drone shot, establishing | Bird's eye overview | | POV | POV shot, first person perspective | Subjective view |

Camera Angles

| Angle | Prompt | Effect | |-------|--------|--------| | Eye Level | Eye level shot | Neutral, natural | | Low Angle | Low angle looking up | Powerful, heroic | | High Angle | High angle looking down | Vulnerable, small | | Bird's Eye | Overhead bird's eye view | Map-like overview | | Dutch Angle | Dutch angle, tilted frame | Tension, unease |

Lighting: The Mood Maker

Lighting changes everything. Same scene, different light = different video.

"Harsh side lighting creates dramatic shadows across the subject's face"
"Soft morning light filters through sheer curtains, warm glow"
"Neon lights reflect off wet pavement in pink and blue"
"Golden hour backlighting creates silhouette effect"

Key lighting terms: Golden hour, blue hour, harsh midday sun, rim-lit silhouette, soft diffused light, neon glow, candlelight, moonlight, high contrast, low key, high key.

Lens Terms That Work

| Lens | Effect | Prompt | |------|--------|--------| | 14mm ultra-wide | Dramatic distortion | shot on 14mm, dramatic perspective | | 24mm wide | Expansive scenes | 24mm wide angle, spacious framing | | 35mm normal | Natural look | 35mm lens, natural perspective | | 50mm portrait | Shallow DoF | 50mm, shallow depth of field, bokeh | | 85mm telephoto | Compressed bg | 85mm telephoto, creamy bokeh | | Anamorphic | Cinematic flares | anamorphic lens, horizontal flares |

Style Keywords

| Style | Keywords | |-------|----------| | Cinematic | cinematic, dramatic lighting, film look | | Documentary | documentary style, natural, handheld | | Film Noir | film noir, high contrast, deep shadows | | Vintage | vintage 16mm film grain, warm tones | | Dreamy | dreamy soft focus, ethereal glow | | Anime | anime style, vibrant colors, cel-shaded |

Common Mistakes to Avoid

1. Too vague — "Make a cool video" tells AI nothing
2. Too many actions — Max 1-2 key actions per video
3. No lighting info — Always include at least basic lighting
4. No camera direction — Static videos feel amateur
5. Exact numbers — "10 birds" → "a flock of birds"
6. Contradictory commands — "slow motion" + "time-lapse" = confused AI
7. Over-prompting — Keep under 50 words
8. Multiple complex movements — Combining rotation + zoom = distortion

Multi-Shot Format (Kling O3)

For multi-cut videos:

3s: Wide establishing shot of the landscape
4s: Medium shot of character walking forward
5s: Close-up of their face, emotional expression
3s: Cut to reaction shot of the other person

Rules: minimum 3 seconds per shot, maximum 6 shots, at least 2 shots to activate multi-shot mode.

Quick Reference Checklist

Before generating, check your prompt has:

• Clear subject (3-5 specific details)
• One main action
• Setting with time of day + atmosphere
• Camera movement or angle
• Style/mood (2-3 keywords max)
• Lighting description
• Under 50 words total