Kling Video 3.0 Model User Guide: https://app.klingai.com/global/quickstart/klingai-video-3-model-user-guide 

Master Formula

[Context/Scene] + [Subject & Appearance] + [Action Timeline] + [Camera Movement] + [Audio & Atmosphere] + [Technical Specs]

Scenario A: The Multi-Shot Narrative (15s)

Goal: Creating a coherent 15-second story with distinct beats.

Prompt:

Shot 1 (0-5s): A wide establishing shot of a desolate Mars colony greenhouse. Outside, a red dust storm rages. Inside, a young botanist in a white bio-suit is kneeling, inspecting a small green sprout.
Shot 2 (5-10s): Cut to a macro close-up of the sprout. The botanist's gloved hand gently touches a leaf. The camera tilts up to reveal her face through the helmet visor; she looks hopeful.
Shot 3 (10-15s): Over-the-shoulder shot. She stands up and looks out the reinforced glass window at the storm. She taps her comms device.
Atmosphere: Cinematic sci-fi, cold blue interior lighting vs. harsh orange exterior.
Audio: The low hum of life-support systems, the muffled howling of wind, and the sharp mechanical click of the comms device.

Scenario B: Dialogue & Lip-Sync

Goal: leveraging native audio and emotional voice acting.

Prompt:

A tense negotiation in a high-stakes corporate boardroom.
Character A (CEO): An older man with silver hair and a sharp suit sits at the head of the table. He leans forward, clasps his hands, and speaks in a deep, authoritative, and gravelly voice: "We are not selling the company. Not today, not ever."
Character B (Rival): A younger woman in a red blazer stands up abruptly. She replies in a sharp, fast-paced, and angry tone: "Then you are sinking this ship with everyone on board!"
Camera: Alternating medium shots focusing on the speaker. Shallow depth of field to isolate the emotions.

Scenario C: High-Octane Action (Physics & Camera)

Goal: Testing motion logic and speed.

Prompt:

A high-speed chase through a Tokyo highway tunnel at night.
Subject: A matte black futuristic motorcycle weaving through traffic.
Action: The bike leans dangerously low into a curve, sparks flying from the footpegs grazing the asphalt. The rider looks back for a split second.
Camera: Dynamic FPV drone shot chasing the bike. The camera rolls 360 degrees as it follows the bike through a narrow gap between two trucks.
Tech Specs: Motion blur, 4K resolution, photorealistic, high contrast.

Scenario D: Text Rendering (Ads & Branding)

Goal: Placing legible text in a commercial video.

Prompt:

A slow-motion product shot of a luxury perfume bottle on a velvet turntable. The bottle is made of crystal glass with gold accents.
Text: clearly embossed on the glass label is the word "ETHEREAL" in an elegant serif font.
Lighting: Soft golden hour light beams hitting the glass, creating refractive caustics.
Movement: The bottle slowly rotates 180 degrees, ensuring the text remains stable and readable throughout the rotation.

Scenario E: Subject Consistency (The "Lookbook")

Goal: Keeping a character consistent across different environments.

Prompt:

Subject: A fashion model with platinum blonde bob-cut hair, wearing an avant-garde geometric silver jacket.
Scene 1: She is walking confidently down a busy New York crosswalk. The camera tracks backward in front of her.
Scene 2: Instant transition. The same model, wearing the same silver jacket, is now standing on a snowy mountain peak. She turns her head to smile at the camera.
Consistency: Ensure facial features and the silver jacket details remain identical between scenes.

Tips for Image to Video Prompts

• Use a Kling Prompter: ChatGPT - Kling 1.6 image to video prompt helper
• Or use ChatGPT
  • Create 20 cinematic scenes for diffrent niche of movies. I want some dynamic move, action, slow motion, stunning scenes or people talking. Give me text prompts in here. Follow my guideline to create text prompts: Subject + subject movement + background + movement of the camera.
• Write short and precise prompts with simple words.
• Add the words "Muted colors" to your prompt because Kling tends to make colors very intense.
• Follow the prompt structure and start with the subject first
• Move the relevance slider up  in the settings.
• Use 5s rather than 10s. Think in small steps.
• Combine Camera Movements
  • Camera moves left and zooms in.
  • Camera moves forward and zooms up.
  • Camera orbits right and follows her behind.
• Consistency
  • character: have ChatGPT write a short and concise parapraph about the character (clothes, pose, body features, facial features, emotion)
  • color consistency: to prevent color change include words like: muted colors, low contrast, desaturated colors
• Control the speed of motion of your video by adding words like 'suble motion' or 'slowly'.
• Use start frame and end frames (two images)
  • to include two movements
  • to change the perspective (front view to side view)
  • to move th subject into focus (from far away to close-up)
• Negative prompts:
  • noise, blur, distortion, disfigurement, low resolution, deformation
  • Smiling, laughing, cartoonish, bright colors, low resolution, morphing, blurry text, disfigured hands, extra fingers
  • morphing, erratic fluctuation in motion, noisy, bad quality, distorted, poorly drawn, blurry, grainy, low resolution, oversaturated, lack of detail, inconsistent lighting 
• Use 'static camera' for images that don't change camera angle.
• Upscale your videos with an upscaler like Topaz AI (4 K))