image | Kling AI
Omni Image
Combine faces, outfits, objects, and scenes from multiple photos into one image — reference anything by label and the AI assembles the composition you describe
image
v3 Omni Latest
Omni Image (O1)
Kling AI
Omni Image lets you mix and match elements from different photos in a single generation. Upload a face photo, an outfit photo, and a background — then write a prompt telling the AI exactly what to take from each one.
The magic is in the labeling system. Every image you upload gets a label — image_1, image_2, and so on. Pre-trained elements from your library get labels like object_1, object_2. In your prompt, you reference these labels with triple brackets: Generate a photo of <<<image_1>>> wearing the dress from <<<image_2>>> in the setting of <<<object_1>>>. The AI understands which part of each reference to use based on your instructions.
This is the tool you reach for when standard image generation is not enough — when you need to put a specific person in a specific outfit in a specific place, all sourced from different images. Face swaps, outfit transfers, style mixing, multi-character scenes, and product placement in custom environments all become a single prompt.
The v3 Omni model adds Series mode, generating 2 to 9 linked images that stay visually consistent — perfect for storyboards, photo series, or marketing sets where every shot needs to feel like part of the same shoot.
The magic is in the labeling system. Every image you upload gets a label — image_1, image_2, and so on. Pre-trained elements from your library get labels like object_1, object_2. In your prompt, you reference these labels with triple brackets: Generate a photo of <<<image_1>>> wearing the dress from <<<image_2>>> in the setting of <<<object_1>>>. The AI understands which part of each reference to use based on your instructions.
This is the tool you reach for when standard image generation is not enough — when you need to put a specific person in a specific outfit in a specific place, all sourced from different images. Face swaps, outfit transfers, style mixing, multi-character scenes, and product placement in custom environments all become a single prompt.
The v3 Omni model adds Series mode, generating 2 to 9 linked images that stay visually consistent — perfect for storyboards, photo series, or marketing sets where every shot needs to feel like part of the same shoot.
Available Models
v3 Omni Latest
kling-v3-omni
Latest — 4K resolution, series generation, intelligent aspect ratio.
Omni Image (O1)
kling-image-o1
Advanced reasoning image model. Supports image references.
Best results
Label Everything Clearly
Each uploaded image gets a label — image_1, image_2, etc. Each element gets object_1, object_2. Use these exact labels in your prompt so the AI knows which reference to pull from.
Be Specific About Roles
Tell the AI what each reference contributes. Say the face from image_1, the outfit from image_2, the background from image_3 — not just use all my images together.
High-Quality Reference Photos
Each reference image should be clear and well-lit. The AI can only extract what it can see — blurry or dark references produce blurry or dark results.
Create Elements for Repeat Characters
If you use the same character often, train them as an Element first. Elements give more consistent identity preservation than uploading photos each time.
Stay Within 10 References
You can use up to 10 references total — images and elements combined. More references means more instructions for the AI to follow, so keep it focused on what matters.
Use Series for Linked Sets
Series mode on v3 Omni generates 2 to 9 images that share the same visual style — ideal for creating content sets, storyboards, or carousel posts that feel cohesive.
Guides
Omni Image
🖼️
Image
Omni Image — Technical Guide
Combine faces, outfits, objects, and scenes from multiple photos into one image — reference anything by label and the AI assembles the composition you describe