klingai-style-transfer

Kling AI Style Transfer & Effects

Overview

Apply artistic styles through prompt engineering, use the Effects API for pre-built visual transformations, and leverage Kolors for image-based style references. Available on v1.6+ models.

Style via Prompt Engineering

The most direct approach -- include style descriptors in your prompt:

import jwt, time, os, requests

BASE = "https://api.klingai.com/v1"

def get_headers():
    ak, sk = os.environ["KLING_ACCESS_KEY"], os.environ["KLING_SECRET_KEY"]
    token = jwt.encode(
        {"iss": ak, "exp": int(time.time()) + 1800, "nbf": int(time.time()) - 5},
        sk, algorithm="HS256", headers={"alg": "HS256", "typ": "JWT"}
    )
    return {"Authorization": f"Bearer {token}", "Content-Type": "application/json"}

# Style: Studio Ghibli watercolor
response = requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
    "model_name": "kling-v2-6",
    "prompt": "A cozy cottage in a meadow, hand-painted watercolor style, "
              "soft pastel colors, Studio Ghibli aesthetic, gentle breeze",
    "negative_prompt": "photorealistic, harsh lighting, dark, gritty",
    "duration": "5",
    "mode": "professional",
    "cfg_scale": 0.7,  # higher = stricter prompt adherence
})

Style Prompt Recipes

Effects API

The Effects API applies pre-built transformations to existing images. Available on v1.6+.

Endpoint: POST https://api.klingai.com/v1/videos/effects

# Apply an effect to an image
response = requests.post(f"{BASE}/videos/effects", headers=get_headers(), json={
    "model_name": "kling-v1-6",
    "image": "https://example.com/portrait.jpg",
    "effect_type": "hug",           # effect to apply
    "duration": "5",
    "mode": "standard",
})

task_id = response.json()["data"]["task_id"]
# Poll for result as usual

Available Effects

Kolors Image Restyle

Use Kolors to restyle images before converting to video:

# Generate styled image with Kolors
image_response = requests.post(f"{BASE}/images/kolors", headers=get_headers(), json={
    "prompt": "A cyberpunk city street, neon signs, rain-slicked roads",
    "aspect_ratio": "16:9",
    "imageCount": 1,
})

# Then use the generated image as input for I2V
image_url = image_response.json()["data"]["images"][0]["url"]
video_response = requests.post(f"{BASE}/videos/image2video", headers=get_headers(), json={
    "model_name": "kling-v2-1",
    "image": image_url,
    "prompt": "Camera slowly pushes forward through the rain, neon reflections",
    "duration": "5",
    "mode": "professional",
})

cfg_scale Tuning

The cfg_scale parameter (0.0-1.0) controls how strictly the model follows your prompt:

For style transfer: Use 0.6-0.8 to ensure the style keywords are respected.

Style Consistency Across Clips

# Use a consistent style template for all clips in a project
STYLE_TEMPLATE = {
    "suffix": ", cinematic lighting, 35mm film grain, warm color grading, "
              "anamorphic lens flare, shallow depth of field",
    "negative": "cartoon, anime, painting, illustration, CGI, digital art",
    "cfg_scale": 0.6,
    "model": "kling-v2-6",
    "mode": "professional",
}

def styled_generation(scene_prompt: str):
    return requests.post(f"{BASE}/videos/text2video", headers=get_headers(), json={
        "model_name": STYLE_TEMPLATE["model"],
        "prompt": scene_prompt + STYLE_TEMPLATE["suffix"],
        "negative_prompt": STYLE_TEMPLATE["negative"],
        "cfg_scale": STYLE_TEMPLATE["cfg_scale"],
        "duration": "5",
        "mode": STYLE_TEMPLATE["mode"],
    })

Resources

• Effects API
• Kolors API
• Developer Portal