Kling O3 AI video generator
Omni model with first and last frame animation and references
- First and last frame
- Multi-reference
- Native audio
JPG, PNG, WEBP, MP4, MOV, MP3 or WAV. Videos and audio 2 to 15 seconds.
Mention files in the prompt as @Image1, @Video1 or @Audio1.
- Give each file one clear job: a face, an outfit, a location or a camera move.
- Mention files in order, like @Image1 and @Video1, so the model knows what to copy from where.
- Reference videos and audio work best between 2 and 15 seconds.
- Audio needs at least one image or video in the stack.
The camera slowly pans upwards as the city awakens, capturing the gradual shift from artificial to natural light, evoking tranquility and wonder.
The camera tilts down to the steaming coffee as rays of sunlight dance around, creating an inviting and warm atmosphere.
The camera slowly zooms in on the figure, who turns and walks deeper into the shadows, heightening the suspense.
The camera circles the DJ as the lights pulse to the beat, capturing the electrifying, energetic ambiance of the dance floor.
The gadget rotates slowly, highlighting its cutting-edge design and technology while the camera gently zooms in to focus on its features.
The camera shakes slightly as hands organize the desk, placing items neatly and opening the notebook, creating an inspiring and relatable scene.
Text overlays appear with tips as the camera smoothly pans across the plants, enhancing the educational and calm tone.
The colors pulse and shift rhythmically as the text pops in and out, creating an exciting and inviting introduction.
The camera glides through the forest, revealing hidden creatures and ancient ruins, building excitement and mystery.
The camera does quick cuts between different poses, capturing the energy and flair of urban fashion.
The camera dramatically swoops over the armies, building anticipation with the swell of the storm and distant battle cries.
The camera captures the motion of a squat with fluid movements, emphasizing form and guidance with upbeat energy.
What is Kling O3?
Kling O3 is the omni model of the Kling family on NovaMotions AI. It combines text, a first and last frame, and up to 7 reference images plus 1 video clip in a single generation, so you can pin down characters, products and motion at once. Clips run 5 or 10 seconds in 720p or 1080p, with native audio when you switch it on.
- longest clip
- 10 s
- highest resolution
- 1080p
- reference files per clip, images and clips mixed
- 8
What Kling O3 does well
The strengths worth knowing before you write your first prompt.
Images plus a motion clip
Add up to 7 reference images and 1 video clip. Use the images for faces, outfits and places, and the clip for the movement or camera path to copy.
First and last frame
Animate a still and set the ending frame. Kling O3 plans the motion between them, which keeps transformations and reveals on target.
Several subjects in one shot
Put two characters and a product into the same scene from separate reference images and describe how they interact.
Optional native audio
Switch on audio for dialogue, effects and ambience generated with the picture. Leave it off when the clip will be scored later.
Social and widescreen formats
Render 5 or 10 second clips in 16:9, 9:16 or 1:1, at 720p or 1080p, ready for feeds, stories and presentations.
What you can make with Kling O3
Product in a lifestyle scene
Combine a packshot, a model photo and a location image so the product appears naturally in use, with the label intact.
Motion transfer
Upload a short clip of a gesture or dance as @Video1 and apply that movement to your own character from @Image1.
Story beats with a cast
Keep two or three recurring characters on model across a series of short clips by reusing the same reference images.
How to use Kling O3
Four steps from an idea to a finished clip, all in the workspace above.
- 01
Choose how to start
Pick Media to build from reference files, Image to animate a first frame, or Text to describe the whole clip in words. You can switch modes without losing your prompt.
- 02
Add frames or references
In Media mode, upload up to 7 images and 1 video clip and give each file one clear job. In Image mode, add a first frame and, if you like, a last frame. Text mode needs nothing.
- 03
Write the prompt
Describe the subject, the action, the setting, the camera move and the mood. Point to your files with @Image1 or @Video1 and put spoken lines in double quotes.
- 04
Set the shot and generate
Choose 5 or 10 seconds, 720p or 1080p, 16:9, 9:16 or 1:1, and audio, check the credit cost on the button, then generate. Keep the best take or extend it with Continue from last frame.
Kling O3 vs Kling 3 vs Seedance 2.0 Pro
Kling O3 is the Kling to use when you have reference files. Kling 3 skips references for longer, 4K cinematic shots, and Seedance 2.0 Pro takes the most references, including audio.
| Compare | This modelKling O3 | Kling 3 | Seedance 2.0 Pro |
|---|---|---|---|
| Best for | Mixed image and clip references | Cinematic 4K shots from frames | The largest reference stacks |
| Ways to create | Media, Image, Text | Image, Text | Media, Image, Text |
| Longest clip | 10 s | 15 s | 15 s |
| Highest resolution | 1080p | 4K | 4K |
| Aspect ratios | 16:9, 9:16, 1:1 | 16:9, 9:16, 1:1 | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 |
| Media references | Up to 7 images and 1 video clip | Not available | Up to 9 images, 3 video clips, and 3 audio files |
| Native audio | Optional | On by default | On by default |
| Typical render time | About 3 minutes | About 3 minutes | About 3 minutes |
| Starting price | From 46 credits | From 30 credits | From 35 credits |
Kling O3 specs
These limits come straight from the workspace settings, so they always match what you can pick above.
- Ways to create
- Media references, Image to video, and Text to video
- Clip lengths
- 5 or 10 seconds
- Resolutions
- 720p or 1080p
- Aspect ratios
- 16:9, 9:16, or 1:1
- Image input
- First frame plus an optional last frame
- Media references
- Up to 7 images and 1 video clip
- Native audio
- Yes, switch it on per clip
- Prompt length
- Up to 10,000 characters
- Takes per run
- 1 to 4
- Typical render time
- About 3 minutes
- Starting price
- From 46 credits
Prompt examples for Kling O3
Press Try this and the prompt lands in the workspace above, ready to generate or tweak.
- Media references
Two friends, one bench
The man from @Image1 and the woman from @Image2 sit on a park bench in autumn, sharing a thermos of tea. She laughs as leaves fall around them. Slow arc around the bench, golden afternoon light, soft wind and distant children playing.
Try thisUpload your files in Media mode first - Media references
Borrowed dance move
The character from @Image1 performs the dance from @Video1 on a rooftop at sunset, city skyline behind. Keep the full body in frame, steady wide shot with a gentle push-in on the final pose, warm backlight.
Try thisUpload your files in Media mode first - Text to video
Tea ceremony detail
Close-up of a hand whisking bright green matcha in a ceramic bowl, fine foam forming, then a slow tilt up to a quiet tatami room with paper screens glowing. Soft diffused daylight, calm and precise movement.
Kling O3 questions
Short answers about modes, limits, credits and rights.
Ready to try Kling O3?
The workspace is at the top of this page. Generations start from 46 credits, and failed jobs return their credits automatically.
Still have a question?
Our team answers within one business day. The prompt guide covers references, camera moves, text and sound.
What is Kling O3?
Kling O3 is an omni video model on NovaMotions AI. It creates 5 or 10 second clips in 720p or 1080p from text, from a first and last frame, or from up to 7 reference images and 1 video clip, with optional native audio.
How many references can Kling O3 take?
Up to 7 images and 1 video clip in Media mode. It does not take audio files as references. Mention them in the prompt as @Image1 to @Image7 and @Video1.
What is the video reference used for?
Use it to copy movement, like a dance, a gesture or a camera path. Describe what to take from it, for example follow the camera move from @Video1, and let the images decide who appears.
Which lengths, resolutions and ratios are available?
Clips are 5 or 10 seconds long, in 720p or 1080p, and in 16:9, 9:16 or 1:1.
Does Kling O3 generate sound?
It can. Switch on audio for dialogue, effects and ambience generated with the picture. Audio is off by default and adds a few credits when you turn it on.
How do I keep a product label readable?
Upload a sharp, front-facing product photo as a reference and say keep the label exactly as in @Image1. Avoid fast spins, a slow orbit or push-in keeps text stable.
Should I use Kling O3 or Kling 3?
Use Kling O3 when you have references to match. Use Kling 3 when you work from text or frames and want clips up to 15 seconds and 4K.
How many credits does a Kling O3 clip use?
It depends on the clip. Longer clips, higher resolutions and native audio use more credits. The exact amount shows on the Generate button before you spend anything, and the same balance works for video, images and music.
What happens if a Kling O3 generation fails?
Every credit goes straight back to your balance automatically. There is nothing to claim or report. Adjust the prompt or the settings and try again.
Can I use Kling O3 videos commercially?
Yes. Every plan includes commercial use and downloads have no watermark. You are still responsible for what you make, so avoid real people, logos or characters you do not have the rights to.
Explore more models
One credit balance works for every model. Switch any time from the sidebar.