Clipset
What is a video combinator?
A video combinator is software that takes a set of modular video clips, usually hooks, bodies, and CTAs, and automatically renders every possible combination of them as separate, finished video files. Instead of editing each ad by hand, you drop in your clips once and the tool assembles all the permutations. It is sometimes called an ad combinator, a video ad combinator, or a hook body CTA combinator. The category exists because paid social teams test lots of creative, and manual assembly is the part that does not scale.
Below: why the math makes it useful, how a combinator differs from tools that sound similar, the two ways these tools are built, and whether you need one.
The math: why 10 × 5 × 3 = 150 matters
Video ads built for testing are usually structured in three segments (the Hook, Body, CTA framework): a hook (the first one to three seconds), a body (the problem, demo, or proof), and a CTA (what to do next).
If each segment is self-contained, any hook can precede any body, and any body can lead into any CTA. The number of finished ads is the product of the counts:
| Hooks | Bodies | CTAs | Finished ads |
|---|---|---|---|
| 3 | 1 | 1 | 3 |
| 5 | 2 | 2 | 20 |
| 10 | 3 | 2 | 60 |
| 10 | 5 | 3 | 150 |
| 20 | 5 | 4 | 400 |
Recording 18 clips (10 + 5 + 3) is an afternoon. Manually cutting 150 timelines from them is a week, and nobody does it. So teams stop at five or ten variations, which is not enough to learn which hook wins.
A video combinator removes that ceiling. The recording work stays the same; the assembly work drops to roughly zero. You can hold the body constant and run ten hooks against it, or hold the hook constant and see whether a testimonial body beats a demo body. The creative testing framework only works if producing variations is cheap.
How a video combinator differs from similar tools
The term gets confused with three other things. They solve different problems.
Video combinator vs. a video editor
A video editor (Premiere, CapCut, DaVinci, Final Cut) is a timeline. You place clips in order, adjust them, and export one file. To make 150 variations you would build 150 timelines and export each one. Editors are built for crafting a single piece, not for enumerating permutations.
A combinator has slots (hook, body, CTA) and a list of clips per slot. It does the ordering and rendering for you. You still use an editor to make the clips; the combinator is what you use after that. See Clipset vs. CapCut for ad variations.
Video combinator vs. Meta Dynamic Creative
Meta’s Dynamic Creative mixes assets inside the ad platform: multiple videos, headlines, primary texts, and calls to action. But it swaps whole videos, not segments within a video. It will not splice hook A onto body B. If you want to test hooks inside the video, each one has to already be rendered as its own file.
A combinator produces those files. The two are complementary. Details in Clipset vs. Meta Dynamic Creative.
Video combinator vs. AI ad generators
AI ad generators create footage: synthetic presenters, generated b-roll, scripted voiceovers. A combinator does not generate anything. It works with clips you already recorded, whether UGC from creators, in-house footage, or existing ad cuts. If your problem is “I have no footage,” you want a generator. If it is “I have footage and need 100 variations of it,” you want a combinator.
Two architectures: cloud vs. local
Every video combinator on the market does the same core job, but they are built two different ways, and that shapes pricing and workflow.
Cloud combinators
You upload clips to a web app, the vendor’s servers render the output, and you download it. Because rendering costs the vendor money, cloud tools are sold as subscriptions and usually meter renders per month. Examples as of September 2026:
- Sovran: cloud subscription, $99 to $399 per month, with 50, 150, or 450 renders per month depending on tier. Also includes AI voiceovers, captions, a timeline editor, and team seats.
- HookScale: cloud subscription, pricing not public. Offers a Video Combinator alongside an AI image generator, ZIP export, and 9:16, 1:1, and 4:5 presets.
- InfiniteAds: cloud, pricing not public. Free first combination, automatic 9:16 formatting.
The cloud model makes sense if you want the extras (captions, voiceovers, collaboration) and your monthly volume fits inside the render cap.
Local desktop combinators
The app runs on your own machine and renders with your own CPU/GPU. Nothing is uploaded, there is no per-render cost, and the vendor has no reason to meter output.
Clipset is the local option in this category: a macOS and Windows desktop app, $49 one-time, free trial, unlimited renders, runs entirely offline. Drop in hooks, bodies, and CTAs (or custom sections), pick any aspect ratio, and it writes every combination as a separate file plus a CSV manifest of what went into each. It does not generate AI footage, add captions, or edit on a timeline. It only combines.
Comparison
| Cloud combinator | Local desktop combinator | |
|---|---|---|
| Where rendering happens | Vendor’s servers | Your computer |
| Pricing model | Monthly subscription | One-time purchase (Clipset: $49) |
| Render limits | Usually capped per month (e.g. 50 to 450) | Unlimited |
| Upload required | Yes | No; clips stay on disk |
| Works offline | No | Yes |
| Render speed | Depends on vendor queue | Depends on your hardware |
| Extras (captions, AI voice, editor) | Often bundled | Typically not; combine only |
| Team collaboration | Often built in | Share output files manually |
| Best fit | Teams wanting an all-in-one cloud suite within a render budget | Individuals and teams rendering high volumes, or who want to keep footage local |
Side-by-side pages: Clipset vs. Sovran, Clipset vs. HookScale, Clipset vs. InfiniteAds.
When you need a video combinator
You probably need one if any of these sound familiar:
- You stop at five variations because assembly is tedious. More hooks would teach you more, but the export grind wins.
- You keep re-exporting the same clips. The body has not changed in a month, yet it gets re-rendered inside every new ad.
- You test hooks. If “which opener works” is a question you ask, you are already thinking in segments. A combinator turns that into a batch job.
- You work with UGC creators. Creators deliver batches of hooks and bodies; a UGC video combiner turns those into a full ad set without a round trip to an editor.
- You run frequent creative refreshes. Weekly test sets are easier when the only new work is recording clips.
When you don’t
The tool is only useful for one job, so be honest here:
- You only need a handful of combos. Three hooks on one body is three edits in any editor.
- You need footage generated. A combinator has nothing to combine until you have clips.
- Your ads are not modular. A single continuous narrative has no segments to swap. You would have to re-shoot modularly first.
- You need heavy per-ad editing. Custom captions, motion graphics, and color work per variation are editor tasks.
How to prepare clips so they combine cleanly
A combinator concatenates. It cannot fix clips that do not fit together. A few habits make the output look intentional rather than stitched:
- Make every segment self-contained. No hook should reference “the thing I just showed you,” and no body should assume a particular opener. The modular UGC script template is built around this.
- Match framing and aspect ratio. Same camera distance, orientation, and resolution. If one hook is 9:16 and the body is 1:1, something gets cropped or padded.
- Match audio levels and room tone. Same mic, same space, normalize loudness.
- Trim tight. Start each clip on the first word and end on the last. Dead air at the seams reads as a mistake.
-
Keep a naming convention.
hook-01-price-objection.mp4,body-02-demo.mp4. With 150 outputs, the manifest that maps each file back to its parts is what makes reporting usable. - Record more hooks than anything else. Hooks vary most in performance, so that is where extra permutations pay off. See how many hooks to test per ad.
Once the clips are clean, the combinator step is mechanical: point it at the folders, choose the aspect ratio, render, upload.
Frequently Asked Questions
Is a video combinator the same as an ad combinator?
Yes, in practice. “Ad combinator,” “video ad combinator,” and “hook body CTA combinator” all describe the same class of tool. “Video combinator” is the term HookScale and InfiniteAds use, and it is the most common label for the category.
Do I need a video combinator if I use Meta Dynamic Creative?
Only if you want to test elements inside the video. Dynamic Creative swaps whole videos, headlines, and text; it does not splice a new hook onto an existing body. A combinator produces the individual files that you then feed to Dynamic Creative or run as separate ads.
How many combinations is too many?
Whatever you cannot get meaningful data on. 150 files is trivial to render but not trivial to test, since each one needs enough spend to say anything. Most teams render the full set, launch a subset (for example, all hooks against one body), and hold the rest for the next round. The bottleneck moves from production to budget, which is where it should be.
Can a video combinator add captions or music?
Some cloud tools bundle captions and AI voiceover (Sovran does). A pure combinator like Clipset does not; it assumes captions are already burned into the clips or added at the platform level. Music is best handled the same way.
Is a local combinator slower than a cloud one?
It depends on your machine and the vendor’s queue. There is no upload or download step locally. The practical difference is not speed but limits: local rendering has no monthly cap, so a 400-ad batch costs the same as a 4-ad batch.