Kling O1 is a creative engine that integrates generation and editing into a single, unified workflow. Unlike previous models where you had to switch between different tools for different tasks, Kling O1 allows you to "Input Anything" (text, images, video, elements) to generate high-quality video content with industrial-grade consistency.
Technical Specifications
To get the best results, ensure your inputs meet these criteria:
Images: Upload up to 4 images. Min resolution 300px, Max file size 10MB.
Videos: Duration 5s–10s. Max resolution 2K. Max file size 200MB.
Output Duration: You can choose between 5s and 10s generations for better pacing control
Common Use Cases
Marketing: Upload product images and a simple prompt to generate high-quality product showcases without an expensive shoot.
Fashion: Create a "virtual runway" by uploading model photos and clothing images. The model retains fabric details perfectly while animating the walk.
Post-Production: Rapidly fix footage (e.g., removing a boom mic or changing a gray sky to blue) using simple text commands, saving hours of manual editing.
How to Use Kling O1 in DFIRST AI
The Kling O1 model in DFIRST is divided into four specialized nodes: Reference Image, Reference Video, Edit Video, and First & Last Images.
Here is a breakdown of each node and what it delivers.
1. First & Last Images
Best for: Controlled transitions and specific narrative arcs.
This node gives you control over the beginning and the end of your clip. The AI acts as the "bridge," calculating the frames needed to get from Point A to Point B smoothly.
Key Inputs:
prompt: Describe what happens between the two images.start_image_url: The very first frame of the video.end_image_url: The very last frame of the video.
Use Case Example:
Set start_image_url as a man standing at a door, and end_image_url as the man sitting inside a car. The model will generate the motion of him opening the door and getting in.
2. Reference Image
Best for: Bringing static characters or products to life with high consistency.
This node focuses on generating video from static visual inputs. It allows you to upload a main image and support it with reference images to ensure the subject (like a specific character or product) looks exactly the same throughout the generated video.
Key Inputs:
prompt: Describe the action, movement, and scene.image_url: The primary image you want to animate.reference_image_urls: Upload up to 4 additional images of your subject (different angles/poses). The model uses these to "lock in" the character's identity.frontal_image_url: Optional specific reference for face consistency.
Use Case Example:
Upload a photo of a fashion model in image_url and three other angles of the same model in reference_image_urls. Use the prompt "Model walks down a cyberpunk runway" to animate them without their face morphing into someone else.
3. Edit Video
Best for: Modifying existing footage (Inpainting/Outpainting) and changing styles.
This is your post-production powerhouse. This node allows you to take an existing video and change specific elements—like the weather, the background, or the clothes a character is wearing—without re-shooting.
Goal | Prompt Structure |
Add Content |
|
Remove Content |
|
Modify Subject |
|
Change Background |
|
Restyle Video |
|
Key Inputs:
prompt: Instructions for the edit (e.g., "Change the dog to a robotic wolf").video_url: The original video clip you want to edit.image_url: (Optional) An image to use as a style or subject reference for the edit.reference_image_urls: Additional references to help the model understand the new look or subject you are inserting.
Use Case Example:
Upload a video of a street in video_url. In the prompt, type "Change the background to a snowy winter day." The cars and camera movement will remain, but the season will change.
4. Reference Video
Best for: Extending storylines or copying camera movements.
Use this node when you already have a video clip and want to generate the next scene in the story, or if you want to apply the specific camera movement of an existing video to a new scene.
Control how your story unfolds by referencing previous clips or specific camera moves.
Generate Next Shot:
Based on [@Video], generate the next shot: [describe content]Match Camera Movement:
Take [@Image] as the start frame. Generate a new video following the camera movement of the [@Video]
Key Inputs:
prompt: Describe the new shot you want to generate.video_url: The source video to use as a reference for context or motion.reference_image_urls: Images to define the visual style or characters for the new shot.
Use Case Example:
Upload a clip of a character looking surprised in video_url. Use the prompt "Generate the next shot: reveal what the character is looking at, a giant dragon." The model uses the context of the first video to generate a coherent sequel shot.
Quick reference: Which node do I need?
I want to... | Use this Node |
Animate a photo of a person or product | Reference Image |
Keep a character consistent across clips | Reference Image |
Copy the camera movement from a clip | Reference Video |
Make a static character dance like a video | Reference Video |
Remove an object from a video | Edit Video |
Change the style of a video (e.g. to Anime) | Edit Video |
Connect two images smoothly | First & Last Images |
🎬 Prompt cheat sheet
Use these formulas to modify existing footage.
Goal | Prompt Formula |
Add Object |
|
Remove Object |
|
Swap Subject |
|
Change BG |
|
Restyle |
|
Recolor |
|
Weather/Env |
|
Pro-tip for all nodes 💡
Reference Image URLs are the secret weapon in Kling O1. In previous models, AI would often "forget" what a character looked like when they turned their head. By uploading up to 4 reference images of your subject (from different angles) into the Reference Image URLs field, you ensure Industrial-Grade Consistency for your characters and props.