How to make a fast-cut meme video for TikTok

Last updated 2026-08-16 · 5 min read

What is the short version?

This format is built from parts everybody recognises: gameplay on one half, oddly satisfying clips on the other, a voice reading something over the top, captions that keep pace. Making one is not creative work so much as sourcing work, and sourcing is where the licensing problems live.

1. Go to Vizard Agent and describe the format and the topic.
2. Say the length and the feed it is for.
3. Say what the voice should be reading, or let Vizard Agent write it.

What do you need before you start?

A format and a script, or at least a topic. Vizard Agent finds the visual layers itself, so what you supply is the thing that makes your video different from the thousand other videos using the same footage — which is the words, almost always.

• The format, described. "Gameplay on the bottom, satisfying clips on top, captions in the middle" is a complete brief.
• The topic or script. This is the only part that is actually yours. A generic script gives a generic video.
• The length. Sixty seconds is standard for this format and it needs a script that fills it.
• The feed. Vertical, and say whether captions should be word-by-word or full lines.
• Any footage you already have. Vizard Agent will use yours before searching, which sidesteps the licensing question entirely.

What do you type into Vizard Agent?

Describe the stack rather than the feeling. This format is layered rather than sequential, so saying what sits where gets Vizard Agent much closer than any adjective would, and it handles the sourcing for each layer once it knows the shape you are building.

Prompt

Variants worth knowing:

• Your own gameplay. Upload a recording and Vizard Agent uses yours as the base layer.
• Different satisfying layer. Soap cutting, slime, sand, hydraulic press. Name it or leave it to Vizard Agent.
• Text only, no voice. Common for feeds watched muted, and cheaper to produce.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real one of these. Almost the entire run is sourcing rather than editing, and the interesting part is watching the searches get more specific as the obvious ones come back with nothing usable attached to them.

1. Lists the asset tools with their capabilities and prices.
2. Checks the download and stock tools and their parameters.
3. Searches the stock library for gameplay parkour, endless-runner footage, soap cutting and slime.
4. Searches YouTube Shorts for the same subjects when stock does not have them.
5. Searches specifically for no-copyright vertical gameplay footage, narrowing the criteria rather than widening the net.
6. Searches for exact watchable URLs for each of the gameplay types.
7. Follows search-result redirects to resolve the actual video links.
8. Searches the stock library again for soap cutting specifically, then YouTube for vertical ASMR versions.

Step 5 is the one that matters. Vizard Agent narrowed its search to material that is licensed for reuse rather than taking whatever matched visually, which is the difference between a video you can post and one that gets claimed.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 60fps, 57.58 seconds, AAC audio. Vertical, just under a minute, at 60fps — which suits gameplay footage and is worth keeping rather than normalising to 30.

Ask for a shorter cut for a different feed and Vizard Agent re-cuts from layers it has already gathered rather than sourcing again.

How long it takes was not measured separately here. Comparable work runs a median of 28 to 38 minutes end to end, and the upper quartile is several times the median. Across all projects the median cost by tier is Flash 47, Pro 55, Max 242, Ultra 263 credits.

When does this not work well?

The visual layers of this format are commodities, and Vizard Agent can assemble them all day. What determines whether the video works is the script, and what determines whether you can keep posting them is where the footage came from.

• Copyright follows the footage. Gameplay, music and clips lifted from elsewhere can be claimed. Vizard Agent searches for reusable material, and the responsibility for what you post remains yours.
• A weak script gives a weak video. The footage is the same footage everyone uses. Only the words are yours.
• Platforms are increasingly hostile to this format. Feeds have started demoting mass-produced content of exactly this shape. It works until it does not.
• Music is the usual claim trigger. Ask Vizard Agent what the bed is, and swap it if the account matters to you.
• The pace is part of the format, not a bonus. These videos cut every second or two by design, and asking Vizard Agent to slow it down produces something that is neither this format nor a good version of anything else.
• "Make it viral" is not an instruction. It gives Vizard Agent nothing to aim at. Give it a topic with something surprising in it.

How do you fix a result that came back wrong?

Name the layer that is wrong. Vizard Agent keeps the sourced clips, the written script and the assembly as separate pieces, so swapping the background footage leaves the captions where they are, and rewriting one line does not send Vizard Agent back out to source the video again.

• "Different gameplay footage." Swapped from the candidates already gathered.
• "Captions are too fast." A timing change on that layer only.
• "Rewrite the third line." The voice is regenerated for that line.

How does Vizard Agent compare to doing it yourself?

The manual version is four browser tabs: one for gameplay, one for satisfying clips, one for a text-to-speech tool, one for a captioning app. Then you stack them in an editor. It takes an hour the first time and forty minutes forever after, per video.

Common questions

Can I use my own gameplay recording? Yes, and it is the cleanest option. Upload it and Vizard Agent uses yours as the base layer.

Will the footage be safe to post? Vizard Agent searches for reusable material, which reduces the risk rather than eliminating it. What you post is your call.

Does it write the script? It can. Give it a topic and Vizard Agent writes and narrates, or paste your own words.

Can it use a different pairing of layers? Yes. Name any two and Vizard Agent will source both, though the more unusual the pairing the more searching it takes to find material that is free to reuse.

How long should the script be? Roughly a hundred and fifty words fills sixty seconds at the pace this format runs. Vizard Agent will fit whatever you give it, and a short script leaves silence the footage cannot carry.

Can it make a batch? Yes. Approve one, then each additional video is a short prompt, because the format is already settled.