Crowds and multiple subjects
The matte is built around a subject. With several people at similar depth the boundary becomes ambiguous and the caption pops in front of someone it should be behind.
Caption style
Captions that go behind the speaker.
The one caption effect that cannot be faked with a font choice. The word has to be occluded by the person in front of it, frame by frame, as they move — which means the tool has to actually know where they are.
Butter Lock — A butter-yellow hero word springs in behind you, white subtitles in front.
Closest alternatives — point at one to play it
Neon Lock
Crimson Lock
Deep Focus
Segmentation runs on your footage in the preview, so you can see it hold before exporting.
What the preset actually is
- Family
- Behind the Person — thirty-odd presets built on the same depth pass.
- Occlusion
- A per-frame subject matte. The caption is drawn, then the subject is drawn over it.
- Motion
- None on the type. The depth is the effect; movement would fight it.
- Preset shown
- Butter Lock. Neon Lock and Crimson Lock are the same build in other palettes.
- Where it works
- Footage with a clear single subject against a distinguishable background.
Why this one is different from the rest
Every other style on this site is a typographic choice: pick a face, a colour, a motion, and it renders the same way over any footage. This one is not a typographic choice at all. It is a compositing operation that depends on what is actually in the frame, and it either holds or it does not.
That is why it is worth doing properly. The effect is widely copied by hand — editors cut the subject out on a few key frames and hope the gap is short enough that nobody notices the edges swim. For a static shot that works. For anything where the speaker moves, gestures, or steps toward camera, hand-masking either takes a long time or looks wrong, and both of those are why most captioning tools do not offer this at all.
Because the caption sits behind the person, it also solves the problem every other style has to work around: covering the subject. A caption block placed over a speaker's chest is a compromise between legibility and hiding them. Behind them, there is nothing to trade off — the text can sit where it composes best.
Where this look is the wrong choice
Low contrast between subject and background
Dark clothing against a dark room gives segmentation very little to work with, and the edge will flicker. Check the preview before committing.
Fast motion and motion blur
A blurred limb has no crisp boundary, so the occlusion edge softens exactly where the eye is looking.
Faceless or screen-recorded content
There is no subject to go behind. Use a normal style — the Faceless preset is built for that footage.
How to put it on your own video
Upload the clip
A single speaker, reasonably separated from the background, works best.
Pick a Behind the Person preset
Butter Lock is the neutral one; the family has thirty more.
Watch the edge, then export
Check the boundary holds while the speaker moves, then export.
Questions about text-behind-subject captions
How does the text get behind the person?
The frame is segmented into subject and background, and that subject matte is composited over the caption. So the caption is genuinely drawn first and the person drawn on top, per frame, rather than a static shape masking a fixed region. It follows the speaker as they move because the matte is recomputed rather than keyframed.
Do I have to mask anything by hand?
No. That is the point of it. The usual way to get this effect is to cut the subject out on key frames in an editor, which is slow and tends to swim at the edges between keys.
Will it work on my footage?
It depends on the shot, and the honest answer is that you should look rather than take a promise. A single speaker with reasonable separation from the background holds well. Crowds, very low contrast and heavy motion blur are the three cases that break it. The preview runs on your own clip, so you can see which you have before exporting anything.
Does the caption still get read if the person covers it?
That is the trade. If the speaker is centred and large, they will occlude a lot of the word, which looks good and reads worse. The presets place type where it usually survives a centred subject, but on a tight close-up you may want a normal style instead.
Read next
Faceless channel captions
The style for footage with no subject to go behind.
Caption safe areas
Where a caption can sit without the interface covering it.
Instagram Reels captions
The same workflow, framed for a Reels edit.
Burned-in vs soft subtitles
A depth effect only exists burned in. What that costs you.
See whether it holds on your footage
Upload a clip and watch the edge while the speaker moves.