Kling Video Edit
Edit existing footage with Kling Video Edit: remove objects, swap backgrounds or restyle a 3–10s clip by prompt. 80 credits/s and up to four reference images.
Kling Video Edit edits footage you already have. Upload a 3–10 second clip, describe the change in plain language, and the model rebuilds the shot at pixel level: removing objects, replacing backgrounds, changing wardrobe, restyling the grade or shifting the time of day. Motion and camera movement stay close to the source. Pricing is 80 credits per second, with an optional set of reference images to guide what the edit should look like.
- Input: Video + prompt
- Duration: 3–10s
- References: Up to 4 images
- Price: 80 credits/s
How to use Kling Video Edit
Use the standard editor for 3–10 second clips when you need prompt-driven changes without the O3 Pro price.
- Upload a clip inside 3–10 seconds Keep the file under 200 MB and at least 700 px wide. Trim to the shot you need, because billing runs on the uploaded duration.
- Pick one thing to change Remove an object, replace the background, change the wardrobe or shift the time of day. One coherent change per run produces the cleanest result.
- Use references for appearance Add up to four JPG or PNG images when the new element must match a specific character, garment or product.
- Decide on the original audio Keep it when dialogue or ambience matters. The model does not generate a replacement soundtrack, so turn it off only if you will add audio in post.
- Check the rebuilt area Compare the edited region frame by frame with the source. Look for texture mismatch and warping at the edges of the change.
Settings reference
Kling Video Edit rebuilds named parts of a 3–10 second clip at 80 credits per second, with optional reference images and original-sound retention.
| Setting | Value | Notes |
|---|---|---|
| Source video | 3–10s | 700–2160 px wide, 200 MB maximum. A prompt is required. |
| Reference images | Up to 4 | JPG or PNG, at least 300 px per side, up to 10 MB each. |
| Price | 80 credits/s | 3s = 240, 5s = 400, 10s = 800. |
| Keep original sound | On / Off | Retains the source audio; the model does not generate a new soundtrack. |
| Output resolution | Follows input | No resolution selector and no resolution-based surcharge. |
| Prompt | Required | Describe the object, background, wardrobe or grade to change. |
What Kling Video Edit changes
This is a semantic editor rather than a filter. It reads the prompt, locates the named subject or region, and regenerates that part of the frame while preserving the rest of the shot. The result is not a mask-and-blur job: removed objects leave plausible background, replaced objects pick up the scene's light and perspective, and style changes propagate across the whole clip.
The model keeps the original timing and camera move, which makes it useful for fixing footage you cannot reshoot. Typical jobs are removing logos or street furniture, swapping a grey sky for golden hour, changing a jacket colour, restyling live action as animation, or replacing a product on a table with a newer model.
- Object removal and clean fill
- Background and environment swaps
- Wardrobe, hair and product replacement
- Style, grade and time-of-day changes
- Original motion and camera preserved
Pricing and the 10-second ceiling
Billing is 80 credits per second of uploaded video, with a 3-second minimum and a 10-second maximum. The upload itself is limited to 3–10 seconds, so the billing cap and the input limit line up: there is no way to submit a longer clip and pay for only part of it.
There is no resolution selector and no separate output-length charge. A 3-second edit costs 240 credits, a 5-second edit 400, and a full 10-second clip 800. Re-running a failed edit costs the same again, so test a difficult change on the shortest clip that shows the problem.
- 80 credits per second
- Minimum charge: 3 seconds
- Maximum: 10 seconds
- No resolution-based surcharge
| Clip length | Credits |
|---|---|
| 3s | 240 |
| 5s | 400 |
| 8s | 640 |
| 10s | 800 |
Writing an edit prompt that lands
Name the target and say what should replace it. “Remove the red car on the left and rebuild the road behind it” works better than “clean up the street”, because the model needs both a location and an end state. Use spatial language — left, right, foreground, behind the subject — and anchor the change to something visible in the frame.
Keep one prompt focused on one coherent change. Two unrelated edits in the same sentence compete for the model's attention and often produce a half-applied result. If you need several changes, run them in sequence, checking each pass before spending on the next. Describe the final look rather than the editing operation: “warm sunset light from the right” beats “change the lighting”.
- Name the object or region to change
- Describe the replacement, not the process
- Use spatial anchors
- One coherent change per run
- Chain edits instead of bundling them
Using reference images
Up to four JPG or PNG images can accompany the prompt. Each must be at least 300px per side and no more than 10MB. References are guidance, not a guaranteed pixel copy: they help the model hold a character's face, a garment's cut or a product's label across the edit, especially when the source clip shows the subject from several angles.
Pick references that match the source clip's lighting and viewpoint. A flat studio product shot is a weak reference for a subject lit by a sunset, and a front-facing portrait does little for a profile shot. When a reference and the prompt disagree, the model has to compromise, so keep the two consistent.
- Up to 4 reference images
- JPG / PNG, ≥300px, ≤10MB each
- Match lighting and angle to the source
- Best for faces, garments and products
- Guidance, not exact compositing
Limits, audio and alternatives
Input clips must be 3–10 seconds, 700–2160px wide and no more than 200MB. A prompt is required. The model does not generate a new soundtrack, but the page lets you keep the original sound, which is usually what you want for dialogue or ambience that should survive the edit.
For longer clips, Kling Video Edit Fast accepts 6–20 seconds at 40 credits per second, at the cost of a longer minimum. The O3 Pro editor handles heavier jobs at 200 credits per second. WAN 2.2 Video Edit is cheaper per second but tops out at 720p, and HappyHorse Video Edit offers 720p and 1080p with its own reference-image workflow.
- 3–10s, width 700–2160px, ≤200MB
- Prompt required
- Original audio can be retained
- No generated soundtrack
- Fast for 6–20s, O3 Pro for heavier edits
Prompt examples for Kling Video Edit
- Remove a logo: Remove the white logo from the jacket on the right side of the frame and rebuild the fabric texture underneath, keeping the original motion and lighting. Name the surface so the fill has something to reconstruct.
- Swap the background: Replace the grey studio background with a sunlit beach at late afternoon, match the subject's rim light, keep the camera movement unchanged. Match the new light direction to the subject.
- Change time of day: Change the scene from midday to blue hour, add warm window light from the left, keep the street reflections and the actor's motion unchanged. One lighting change per run.
- Wardrobe replacement: Replace the man's black t-shirt with a charcoal knit sweater, keep the collar shape and the way the fabric moves with his walk. Reference images help hold the garment cut.
- Product swap: Replace the white bottle on the table with a matte black bottle, keep the label position, reflections and the slow push-in shot. Use a product reference for label accuracy.
- Style transfer: Restyle the footage as hand-painted 2D animation with clean outlines and flat colour, preserve the actor's performance and camera framing. Style edits read better on shorter clips.
Pro tips for Kling Video Edit
The standard editor is half the price of O3 Pro and handles most object, background and wardrobe changes.
- Choose the editor by clip length Standard handles 3–10 seconds; Fast starts at 6 seconds. If your clip is under 6 seconds, this is the only Kling editor that bills it fully.
- Write the target, not the process Say what the shot should show, not how to mask or track it. One sentence naming the region and the result is enough.
- Match the light in the reference A reference image with the same light direction as the source blends faster. Studio-white references on a warm scene look pasted on.
- Keep the source camera The model preserves motion and camera movement. Do not ask it to dolly, pan or cut; change the content inside the existing move.
- Retain audio for talking shots Keeping the original sound avoids a mute clip and keeps lip movement aligned with the existing dialogue.
Frequently asked questions
How much does Kling Video Edit cost?
80 credits per second of uploaded video. A 3-second clip is 240 credits, a 5-second clip 400 and a 10-second clip 800.
How long can the input video be?
3–10 seconds. The same range is used for billing: shorter clips are charged at 3 seconds, and nothing longer than 10 seconds is accepted.
Is a prompt required?
Yes. The model needs a natural-language instruction describing what to change and what the result should look like.
What are the reference images for?
Up to four JPG/PNG images can guide the appearance of a character, garment or product in the edit. They must be at least 300px per side and no more than 10MB each.
Does it keep the original audio?
You can choose to retain the original sound. The model does not generate a new soundtrack; for that, use Add Sound to Video or a dedicated audio model.
What kinds of edits work best?
Object removal, background swaps, wardrobe and product replacement, lighting changes and style transfer. Large changes that alter the subject's identity are harder to keep consistent.
How does it differ from Kling Video Edit Fast?
The standard model handles 3–10 second clips at 80 credits per second. Fast handles 6–20 seconds at 40 credits per second, so it suits longer but simpler edits.
Can I make several edits in one run?
Related changes can be combined, but unrelated edits often compete. For the most predictable result, apply one change per run and check each pass before continuing.
Compare with other AI video editors
- Kling Video Edit Fast
- Kling O3 Pro Video Edit
- WAN 2.2 Video Edit
- HappyHorse 1.0 Video Edit
- Kling Image-to-Video
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Video editing
- Credit pricing: 80 credits per second
- Capabilities: new, popular, editing
About Conut.ai
Conut.ai is an AI image, video and audio generation platform with 50+ creative tools. It supports text-to-image, image-to-video, face swap, motion control, relighting, inpainting and video effects through models including Nano Banana Pro, Seedream, GPT Image, Sora 2, Kling, WAN and Seedance. Start free — no credit card required.
Explore more AI tools
- HappyHorse 1.0 Text-to-Video
- HappyHorse 1.0 Image-to-Video
- HappyHorse 1.0 Video Edit
- HappyHorse 1.0 Video Extend
- Seedance 2.0 Text-to-Video
- Seedance 2.0 Image-to-Video
- Seedance 2.0 Fast Text-to-Video
- Seedance 2.0 Fast Image-to-Video
- Seedance 2.0 Advanced
- Image Watermark Remover
- Image Background Remover
- Video Watermark Remover