Why word-by-word
In OpusClip’s 2026 analysis of high-performing short videos, about 80% of the top tier burned captions into the video and most animated them, usually as single words or short bursts (OpusClip). One word at a time also matches fast narration: at around four words a second, each word is on screen for about a quarter of a second.
The two styles
| Narrator (“creator”) | Skit caps | |
|---|---|---|
| Used for | Narrated promos and showcases | Skits with two or three characters |
| Case | Sentence case, punctuation kept | ALL CAPS, punctuation kept |
| Font | Bold geometric sans (Poppins Bold matches) | Heavy rounded or geometric display font |
| Look | White, soft black outline-glow, soft shadow | White, thick dark outline, hard shadow |
| Position | Centre about 70–72% down the frame | Centre about 28% down (upper third) |
| Size | About 4.6–5% of frame height | About 6–7% of frame height |
| Animation | Hard swap, no fade | Quick pop from about 85% to 100% scale |
| Colour | Red only on the hook’s key word(s) | One colour per speaker; red for key words |
These are measured from reference Minecraft Shorts frame by frame. Both styles keep each word on screen until the next one starts, so there is always a word to read.
Speaker colours in skits
Give every character one caption colour and keep it for the whole video: the narrator white, the second character yellow, the third light blue. Viewers follow who’s talking without seeing faces, which matters when two players look alike.
Placement rules
- Never cover the key action. Skit captions sit high so the characters stay visible; narrator captions sit low over gameplay.
- Stay out of the platform buttons: the bottom quarter and the right-hand column.
- Captions go above game elements such as menus and nametags.
- A counter or label can stack directly above the caption.
Timing
- Each word appears at its spoken start and stays until the next word starts.
- At a sentence break, clearing the caption for 50–150 ms reads as a breath.
- Cuts land on the start of a word, so the caption change and the cut feel like one beat.
Captions in VoxelCut
VoxelCut generates the voice with a timestamp for every word, so captions are timed automatically. Both styles above are built-in presets, with speaker colours and hook emphasis. You can restyle any word in the editor.
Made with VoxelCut · Comparison Short
It’s Clash of Clans, but in Minecraft
One-word captions in “It’s Clash of Clans, but in Minecraft”.
See all examplesFrequently asked questions
What font do Minecraft Shorts use for captions?
Narrated videos typically use a bold geometric sans such as Poppins Bold; skits use heavy display fonts in all caps. Minecraft-style pixel text is used for counters and labels rather than captions.
Where should captions go on a Minecraft Short?
About 70% down the frame for narration, or about 28% down for skits, always clear of the platform’s buttons at the bottom and right.
Should captions be one word at a time?
For fast narrated Shorts, yes. It matches the speech rate, is easy to read on a phone, and is what most top Minecraft Shorts use.
Updated