Reference
OutputThe person. Keep the performance and the framing, use your own talent, founder, model or yourself. This is the version most people start with — same choreography, different face on screen.
CloneCut is an AI video cloner. You send a reference clip and one image, and it rebuilds that clip with your own person, product, object or scene inside it — keeping the motion, the camera path, the cut points and the pacing of the original.
No timeline. No keyframes. No shoot.
Reference
Input image
OutputCloneCut answers one narrow question: how do I make that video, but with my own subject in it?
Most tools answer a broader question — how do I make a video — and expect you to describe the result in words. CloneCut starts from a clip that already works. It reads the clip's structure, then renders a new version of it with the person, product, object or scene you supply. The format survives; the cast changes.
That is the whole product. Everything else on this page is the detail behind that sentence.
CloneCut is a reconstruction tool, not a generator. The distinction matters more than any feature list.
If you already have a reference in mind, CloneCut removes the shoot between watching a format and publishing your own version of it. You can read more about CloneCut and why we built it that way.
Every CloneCut job is defined by two answers: what has to survive, and what you choose to replace.
What stays
What you replace
The left column is not a setting you tune. It is inherited from the reference, because the reference is the input. The right column is a single decision you make before you render.
Pick a short, clean clip — roughly three to twenty seconds, one obvious subject, a camera you can follow. Vertical 9:16 for TikTok, Reels and Shorts works, and so does landscape footage.
Upload a sharp photo of the person, product, object or scene that should appear in the new version. Front-facing, evenly lit, nothing covering it. One good image is enough.
Describe the change in plain words — "keep the same motion and camera, use this product instead." CloneCut renders the clip and hands you an MP4.
That is the entire workflow. If you want the longer version with input examples, see how CloneCut works.
A description of a video tool proves nothing. Watch the reference on the left and the output on the right, on the same timeline, and judge the motion rather than the stills.


Drag to compare
Reference
OutputThe person. Keep the performance and the framing, use your own talent, founder, model or yourself. This is the version most people start with — same choreography, different face on screen.
Reference
OutputThe product. The hand path, the reveal angle, the turn of the wrist and the final hero frame all come from the reference; only the item in the frame changes. Ecommerce teams use this to match a format a competitor is already running.
Reference
OutputThe object. The physical action is what makes the shot work, and the thing being handled is interchangeable — something opened, poured, thrown or worn.
Reference
OutputThe scene. The subject and the performance stay; the environment changes. This is the category with the widest quality spread, because the reference's lighting has to stay plausible for the new location.
When you want a specific format with your own subject in it, there are only four paths. Each has a place; only one of them starts from the clip.
Reshoot
Text-to-video
Template editors
CloneCut
The point of the table is not that the other three are bad. It is that a reshoot rebuilds a format from memory, a prompt guesses at it, and a template never had it. CloneCut is the only path where the format is an input.
It fits you if you already have a reference in mind and want your own version of it — matching a winning product-video format, joining a trend without rebuilding it, producing ad variants without booking a shoot, or running the same format across several clients.
It does not fit if you want a video generated from a script with no source footage, you need a long-form narrative, you cannot supply a usable reference clip, or you need manual control over every element in every frame. Those are different tools.
Be honest about your inputs, too. A clip with one clear subject, even light and a camera you can follow clones well. Fast cuts, several overlapping people, heavy compression and very dark footage are the hard cases — not because the tool refuses them, but because there is less structure to copy.


Resolution
Credit cost
Output is a finished file with no editing layers attached, so it can go straight into a scheduler or an editor.
Three plans, billed monthly or yearly. Yearly billing halves the monthly rate.
Lite — $29/month. $14.50 per month on yearly billing, 96,000 credits per year. For occasional renders and testing a format before committing.
Pro — $59/month. $29.50 per month on yearly billing, 216,000 credits per year. For creators and small teams publishing weekly.
Ultra — $119/month. $59.50 per month on yearly billing, 432,000 credits per year. For agencies running the same format across several clients.
Credit packs are available on top of any plan if you render in bursts. Full plan comparison and current limits are on CloneCut pricing.
CloneCut is built for footage you have the right to use and for people who have agreed to appear in it. Use your own clips, or clips you are licensed to use. Do not place a real, identifiable person into a video without their permission, do not put words in someone's mouth, and do not misrepresent anyone.
Label AI-edited video wherever a platform or local law asks you to. TikTok, YouTube, Meta and most schedulers now carry their own labels for synthetic media, and the label is part of publishing, not an afterthought.
These are conditions of use in our terms, not guidance we work around.
CloneCut is built and operated by the team behind this site. Other websites publish under the same name on different domains; we do not run them, we are not connected to them, and their pricing, output quality and data handling are not ours.
If you want the team that answers effect requests and publishes its own failures, this is it. The CloneCut examples page is the fastest way to tell the difference — the outputs there are ours, unedited, including the ones that did not work.
CloneCut is an AI video cloner. You give it a short reference clip and one image of your own person, product, object or scene, and it renders a new version of that clip with your subject in it, keeping the original motion, camera movement, cut points and pacing. It is not a text-to-video generator — the source footage is the input, not a sentence.
No. The whole workflow is a clip, an image and one sentence describing what to change. There is no timeline, no keyframing, no masking and nothing to install — it runs in the browser on a laptop or a phone.
Three to twenty seconds, one clear subject, even light, and a camera you can follow. Vertical 9:16 for short-form platforms and landscape both work. Fast cuts, several people overlapping in frame, very dark footage and heavily compressed sources give the tool less structure to copy, so results degrade there first.
Yes. A product, an object, a character or a mascot all work, as long as one image shows it clearly. Product clips are the most requested category because the hand path, the reveal angle and the final hero frame all come from the reference while only the item changes.
Up to 20 seconds per render, at 480p, 720p or 1080p, delivered as an MP4 with no watermark. The credit cost of a job is shown before you render, and credits for a failed render are returned.
Plans start at $29/month, or $14.50/month on yearly billing, and go up through Pro and Ultra for heavier rendering. Credit packs are available on top of any plan. There is no free plan or free trial — every render is paid for with credits, and you see the credit cost before you render. Current limits and every line item are on the CloneCut pricing page.
No. CloneCut is built and operated here. Sites publishing under the same name on a different domain are separate products run by other people; their pricing and output are not ours. If you want to confirm you are on the right one, the examples page is the quickest check.
Send one reference clip and one image. Keep the motion, keep the camera, keep the timing — change what is in the frame.