Kling Video 3.0 Omni
A unified multimodal video model that generates and edits video from text, images, reference images, and existing video. Kling Video 3.0 Omni combines text-to-video, image-to-video, reference-based generation, and video editing into a single model with native audio output.
What it does
Kling Video 3.0 Omni takes multiple input types—text prompts, starting images, reference images, and reference videos—and generates cinematic video clips up to 15 seconds at up to 1080p. It maintains character and style consistency across scenes using reference inputs, and can edit existing videos based on text instructions.
The “Omni” in the name reflects its unified approach: instead of separate models for different tasks, this one model handles generation, editing, and style transfer.
How to use it
Text to video
Describe what you want to see. The model generates video from your text prompt.
Image to video
Upload a start_image (and optionally an end_image) with a text prompt describing the motion. The model animates your image into a video.
Reference-based generation
Upload up to 7 reference images to guide the model on character appearance, scene style, or specific elements. In your prompt, use <<<image_1>>>, <<<image_2>>>, etc. to refer to each reference image.
Example prompt with references: <<<image_1>>> walks into a coffee shop and sits down at a table near the window. The interior matches the style of <<<image_2>>>.
Video editing
Upload a reference_video and set video_reference_type to base. Then describe the edits you want in your text prompt—the model will modify the existing video according to your instructions.
Style and camera reference
Upload a reference_video and set video_reference_type to feature. The model will use the video’s camera movement, style, or visual characteristics as a reference for generating new content.
Multi-shot mode
Use the multi_prompt parameter for videos with multiple scenes. Pass a JSON array of shot definitions, each with a prompt and duration (up to 6 shots).
[
{"prompt": "<<<image_1>>> opens a door and steps inside", "duration": 5},
{"prompt": "<<<image_1>>> looks around the room in surprise", "duration": 5}
]
Writing effective prompts
- Scene and motion: Describe what happens, where, and how the camera moves
- Reference tags: Use
<<<image_N>>>or<<<video_1>>>to refer to your uploaded references - Audio (if enabled): Put dialogue in quotation marks, describe ambient sounds explicitly
- Be specific: “A woman in a red coat walks briskly through falling snow” works better than “a person walking”
Parameters
- mode:
standard(720p) orpro(1080p) - duration: 3 to 15 seconds (ignored for video editing)
- aspect_ratio: 16:9, 9:16, or 1:1
- reference_images: Up to 7 reference images (up to 4 when also using a reference video)
- reference_video: A video for style reference or editing (.mp4/.mov, 3-10s, max 200MB)
- video_reference_type:
feature(style/camera reference) orbase(video editing) - generate_audio: Toggle native audio (mutually exclusive with reference video)
- keep_original_sound: Keep the audio from the reference video
- negative_prompt: Describe what to exclude
What it’s good for
- Consistent characters across scenes: Use reference images to keep the same character look throughout a project
- Video editing and restyling: Modify existing footage with text instructions
- Marketing and advertising: Create polished videos with brand-consistent visuals
- Multi-scene narratives: Build cohesive stories with multi-shot mode and reference consistency
- Style transfer: Apply the look of a reference video to new content
Limitations
- Maximum 15 seconds per generation
- Audio generation can’t be used together with reference video input
- Reference video must be between 3 and 10 seconds
- Character consistency improves with clear, well-lit reference images
- Complex physics interactions may not look fully natural
Privacy policy
https://app.klingai.com/global/dev/document-api/protocols/privacyPolicy
API terms
https://app.klingai.com/global/dev/document-api/protocols/paidServiceProtocol
Service level agreement
https://app.klingai.com/global/dev/document-api/protocols/paidLevelProtocol