Does it work with a handheld or moving camera?

Yes. The matting model finds the person in each frame and carries information from one frame to the next, so pans, handheld wobble and dark clothes on a dark wall hold steady instead of flickering.

Can I put text behind an object, like a chair or a car?

No. The matte is a person matte: people, and whatever they carry, pass in front of the words. Furniture, walls and parked cars do not, so over those the words read as an ordinary title. That trade is what keeps the effect steady frame to frame.

Why did Valmera refuse to put the text behind me?

It refuses with the reason so you know what to change: nobody visible in the window, the subject filling more than 55% of the frame, a cut inside the window, a speed change over the footage, a window longer than 15 seconds, or a moment that plays an inserted clip. Nothing is changed, and a plain title is offered instead.

Why does the text come out so large?

Because big type is what reads as depth. With tall letters the person crosses the middle of the word while the tops and bottoms stay visible, so the eye completes it. A small line sits inside the body and whole words vanish. If letters get lost, make the text bigger or shorter, never smaller.

What happens if I cut the video afterwards?

The words stay attached to the footage they were measured on, so earlier cuts move them along with their shot. If you cut that shot out entirely, the text falls back to an ordinary title rather than disappearing; remove or re-place it.

Creating an account and uploading are free. Editing requires a subscription. Exports are full-resolution H.264 MP4s and end with a brief Valmera end card.

Put Text Behind the Subject in a Video

To put text behind a person in a video, upload the footage to Valmera and type “put my name behind me as I walk across at 0:12.” The AI agent cuts the person out of every frame in that window with a video-matting model, draws the words, and lays the person back over the letters, so they pass in front of the title like a sign on the wall behind them.

It is a real depth composite, not a fade or a transparency, and it needs no green screen, rotoscoping or keyframed mask.

Try the Depth Text Effect

Upload footage, type one sentence, watch the preview. Creating an account and uploading are free; editing requires a subscription.

Create an account →

See pricing →

How to Put Text Behind a Subject in Valmera

1
Upload footage with a person in it
MP4, MOV, M4V, MKV or WebM, up to 100 GB and 3 hours. Wider shots where someone crosses the frame work best.
2
Say the words and the moment
“Put VALMERA behind me as I walk across at 0:12, for about 3 seconds,” or describe the shot: “when I step away from the desk.”
3
Read the reply
The agent reports how much of the title the person actually crosses. Near zero means it will look like an ordinary title, so move the text or the window.
4
Adjust and export
Ask for bigger type, a new position or another font, then export a full-resolution MP4 rendered from your original file.

PROMPTS TO PASTE

Put my name behind me as I walk across at 0:12.

The word LAUNCH behind me in Anton, white, for 3 seconds at 0:45.

Move that text left so I cross the middle of it.

Make the behind text bigger. The letters are getting swallowed.

How the Effect Is Built

The mask comes from a video-matting model that carries information from one frame to the next, so the silhouette holds steady instead of flickering at the edges. It is measured on a lightweight copy for speed, but only the mask is stored: the final render draws the words onto its own full-resolution picture and uses the mask to put your pixels back on top, at source quality.

Settings That Matter

Size. Behind-subject titles default to large type, about 2.4 times a normal title. The person should cross the middle of tall letters. If words vanish, go bigger or shorter.
Placement. Put the words where the person sweeps through them once, then let the letters re-emerge. Someone standing still in front of a word just hides it.
Window. 0.4 to 15 seconds, inside one continuous take. Two to four seconds reads well.
Style. The same templates, fonts, colors and entrances as add text to video.
What hides the words. People and what they carry. Furniture, walls and parked cars do not.

When Valmera Refuses, and What to Change

Rather than render a broken version, the agent refuses with the reason, changes nothing, and offers a plain title.

Reason	What it means	What to do
Nobody is in the shot	Under 0.4% of the frame reads as a person	Pick a moment where someone is on camera, or use a plain title.
The subject fills the frame	Over 55% average coverage	Use a wider shot where the person crosses the frame.
A cut inside the window	The window spans two takes	Shorten it or move it inside one shot.
A speed change over it	The footage is sped up or slowed down	Remove the speed change there, or place the words elsewhere.
Too long	Over 15 seconds	Split it into shorter placements.
On an inserted clip	The moment plays b-roll, not your footage	Aim at a moment where your own video is playing.

Common Mistakes

Small type. A small line sits entirely inside the body, so whole words disappear. Big type is the look.
A close-up. When a person fills the frame there is nowhere for the words to live. Use a wider shot.
A zoom or speed change over the window. The mask is tied to the source frames; re-timing or re-framing that footage moves the picture out from under it.
Using it on every title. Once or twice per video it reads as production value; on every title it reads as a template.

Limits

People only: objects do not occlude the words. Each placement lasts up to 15 seconds and must sit inside one take of your own footage, with no speed change over it. Very fast motion and loose, fine hair are the hardest edges for any matte; slower movement across the words looks best. The effect needs a main video, so it is not available in a project built only from images.

When to Use Something Else

After Effects' Roto Brush gives frame-level control and can matte objects as well as people, if you know After Effects. DaVinci Resolve's Magic Mask tracks a subject you mark inside a node workflow. CapCut users stack a background-removed copy of the clip over a text layer by hand. Valmera fits when you want the effect from one sentence, with the rest of the edit in the same chat.