Skip to main content
Unifically LogoUnificAlly
Model logo

Kling O1 Edit API

  • Video to Video
PlaygroundREADME

Upload images/videos above, then type @ in your prompt to reference them

Click or drag & dropMP4, WEBM, MOV · Max 100MB
Click or drag & dropPNG, JPG, WEBP, GIF · Max 100MB
Keep audio
Keep original video audio
Output

Your generated video will appear here

Kling O1 Edit

Unifically: one endpoint, one API key, USD billing.

What is Kling O1 Edit?

Kling O1 Edit is the video-to-video editing route under Kling O1. Pass a source video (up to 10 minutes from the playground) and a prompt describing the edit, and the model returns a new clip of the same length. The reference bundle is up to 4 reference images (image_urls). Two video modes drive the result: Reference uses the source as a style or motion guide while the prompt steers the change; Transform reshapes the source more aggressively. keep_audio preserves the original soundtrack when sound continuity matters. Output renders at Std (720p) or Pro (1080p), in 16:9, 9:16, or 1:1. Pick O1 Edit over Kling 3.0 Omni Edit when the reference shape is purely image-based and you do not need video elements.

Key features of Kling O1 Edit

Four features cover what teams build with O1 Edit.

Edit a source video with a single prompt

Pass video_url and a prompt describing the edit. Output length follows input length, so the timeline stays put while the visual changes.

Reference and Transform video modes

Reference treats the source as a style or motion guide for the prompted change. Transform reshapes the source itself with a heavier restyle or retime.

Up to 4 image references

Pass image_urls (max 4) and reference them inline in the prompt as @image_1 through @image_4. Useful for steering wardrobe, environment, or colour treatment from a still.

Audio passthrough on demand

keep_audio defaults to off. Set it to true to preserve the source soundtrack in the output, which is what you want for ads going to multiple regions with the same VO.

Best for

Plate revisions

Same-length edit passes without rebuilding timeline audio. Useful when the audio mix is final but visuals need a tweak.

Style swaps via image references

Reference images steer wardrobe, environment, or colour treatment from a single still.

Light VFX via Transform mode

Heavier restyle than a colour tweak when the brief calls for it.

Localization with audio preserved

Visual changes (signage, props, environment) while keeping the original soundtrack intact.

Client alternate cuts

Same source, different prompts for client review without re-shooting or NLE work.

Image-only reference pipelines

Predictable image_urls shape keeps the reference contract simple in production code.

Variants

Two output tracks plus two edit modes. Pick by resolution and by how aggressive the edit needs to be.

Std

The 720p output mode. Good for editorial passes, alternate-cut review, and client-facing iteration where the cut is not the delivery yet.

Pro

The 1080p output mode. Use it when the edit is the delivery cut and 1080p is the contract.

Reference mode

Treats the source clip as a style or motion guide for the prompted change. Good for edits that should preserve the original feel: costume swap, lighting shift, signage replacement, environment tweak.

Transform mode

Reshapes the supplied footage more aggressively. Heavier restyle, retime, or scene-level rework. Use it when the edit is a meaningful departure from the source plate.

Use cases

Swap a costume in a finished spot by passing the source video, the wardrobe reference as @image_1, and a Transform-mode prompt that names the change. Localize a global ad for a new region by keeping audio passthrough on while reprompting the visible signage and street furniture in the background. Build three alternate cuts for client review by holding the source constant and varying only the edit prompt and reference image. Apply a light VFX colour and look pass to a trailer by switching to Transform mode without rebuilding the timeline.

API examples

Call Kling O1 Edit from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/video/kling/kling-o1-video-edit.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "kuaishou/kling-o1-video-edit",
    "input": {
      "prompt": "Transform into anime style",
      "video_url": "https://example.com/video.mp4",
      "video_mode": "reference",
      "mode": "pro"
    }
  }'

Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.

FAQs

People also ask

It is the video-to-video editing route under Kling O1. Pass a source video and a prompt describing the edit; the model returns a new clip of the same length. Up to 4 reference images condition the change. Two video modes (Reference, Transform) and an optional keep_audio toggle round out the controls.

Generate creates a fresh clip from a prompt and references. Edit always starts from your footage and applies the described change. Use Generate when you have nothing yet; use Edit when you have a clip that needs revision.

Both reshape an input clip with the same Reference and Transform mode choice. Omni Edit adds an elements parameter that supports both image and video element types. O1 Edit only takes image_urls. Pick O1 Edit when the reference shape is purely image-based.

Source video uploads up to 10 minutes per the playground constraint. Output duration is locked to the input video length and cannot be set manually.

Use @image_1, @image_2, @image_3, @image_4 for entries in the image_urls array. Reference them inline in the prompt string.

Modèles liés

Voir tous les modèles →