kling-v3 - Image to Video

Kuaishou’s Kling 3.0 i2v model. Generate videos from reference images with camera control, motion brush, and multi-shot storyboarding.

Kling 3.0 i2v is an image-to-video generation model from Kuaishou’s Kling series, available through AnyFast API. Transform static images into dynamic videos with customizable camera movements, motion brushes, and multi-shot storyboarding.

This page uses Kling’s Legacy API: POST /kling/v1/videos/image2video with model_name: "kling-3.0". It does not use the upgraded /image-to-video/kling-3.0 request structure.

Key capabilities

• Image-to-Video — Generate videos from first frame and/or end frame reference images
• Camera Control — Preset and custom camera movements (pan, tilt, zoom, roll)
• Motion Brush — Static and dynamic brush masks for precise motion control
• Multi-Shot — Create up to 6 storyboard shots with custom prompts and durations
• Sound and Voice — Generate synchronized sound or reference up to 2 voices
• Multiple Output Modes — std (720p), pro (1080p), and 4k

Quick example

cURL

curl https://www.anyfast.ai/kling/v1/videos/image2video \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model_name": "kling-3.0",
    "image": "https://example.com/photo.jpg",
    "prompt": "The person smiles and waves to the camera",
    "duration": "5",
    "mode": "std"
  }'

Parameters

Parameter	Type	Required	Description
model_name	string	Yes	Must be kling-3.0
image	string	Conditional	First-frame image (raw Base64 or URL)
image_tail	string	Conditional	End-frame image (raw Base64 or URL)
prompt	string	Conditional	Text prompt (max 2500 chars)
negative_prompt	string	No	Negative prompt (max 2500 chars)
mode	string	No	std (720p), pro (1080p), or 4k. Default: std
duration	string	No	3–15 seconds. Default: 5
cfg_scale	float	No	Prompt relevance 0–1. Default: 0.5
multi_shot	boolean	No	Enable multi-shot storyboarding. Default: false
shot_type	string	Conditional	customize or intelligence. Required when multi_shot is true
multi_prompt	array	Conditional	1–6 shots. Required for customized storyboards; shot durations must sum to duration
element_list	array	No	Up to 3 element IDs from the Kling element library
voice_list	array	No	Up to 2 voice IDs. Mutually exclusive with element_list; referenced voices require sound: "on"
camera_control	object	No	Camera movement control
static_mask	string	No	Static motion-brush mask, using the same image rules and aspect ratio as image
dynamic_masks	array	No	Up to 6 dynamic masks with trajectories
sound	string	No	on or off. Default: off
watermark_info	object	No	Set enabled to generate an additional watermarked result

Input rules

• Provide at least one of image or image_tail; last-frame-only generation is supported by this Legacy endpoint when the selected model/mode allows it.
• Images support .jpg, .jpeg, and .png, up to 10 MB, with both dimensions at least 300 px and an aspect ratio from 1:2.5 to 2.5:1. Raw Base64 must not include a data:image/...;base64, prefix.
• image_tail, motion-brush masks, and camera_control are mutually exclusive.
• Each dynamic trajectory contains 2–77 coordinates for a 5-second video. Static and dynamic masks must use the same resolution.
• element_list and voice_list cannot be used together.

API Reference

View the interactive API playground for Kling 3.0 i2v.