Add Text to Video
Type a title, a hook or a credit onto your clip and export the MP4, with no watermark.
- Free
- No sign-up
- No watermark
- Runs in your browser
Loading the tool.
How big, where, and for how long?
Those three questions are the whole job, and only the first one has a wrong answer that is easy to give: a size in pixels. Text set at 48 px is a banner on a 480p clip and a whisper on a 4K one, so here the size is a share of the frame's shorter side, and the same setting looks the same on a phone video and on a wide one. Position and timing are yours to pick; everything else below is what the tool does with what you picked.
| Setting | What you choose | Starts at | On a 1080 × 1920 clip |
|---|---|---|---|
| Position | Top, middle or bottom | Top on vertical video, bottom on wide video | 134 px of clear picture between the block and that edge |
| Text size | 3% to 12% of the shorter side | 6.5% | 70 px letters, 85 px from one line down to the next |
| Color | White, yellow or black letters | White | The backdrop flips to the opposite ink: black behind white and yellow, white behind black |
| Backdrop | An outline, a box, or neither | Outline | A 5.6 px rim around every letter |
| Time on screen | The whole video, or one stretch you drag | The whole video | The handles and both fields step by a tenth of a second |
| Line breaks | Yours are kept, the rest are wrapped for you | Four lines at most | Each line is measured against 870 px of width |
The pixel column is arithmetic on the frame, not a preset. The size is the shorter side multiplied by your percentage: 1080 × 0.065 gives 70 px on a portrait phone clip, and 720 × 0.065 gives 47 px on a 1280 × 720 one, which is why the two look alike side by side. The margin is 7% of the frame height (1920 × 0.07 = 134 px) and the wrapping width is 86% of the frame width less the padding either side (1080 × 0.86 − 58.8 = 870 px). When a block still needs a fifth line, it is set down to 80% of your size — 56 px here — before a single character is dropped, and the workbench then prints that smaller number in place of the one you asked for, because the size in the file is the only one worth reading.
How it works
Open the clip. Drop an MP4, MOV, WebM or MKV onto the box, or pick it from your disk. Its length, its size and its tracks are read on the spot, and the file stays where it is: nothing is sent anywhere, and there is no queue and no sign-in between you and the next screen.
Type it and place it. Write up to 300 characters, keeping your own line breaks, then set the position, the color, the backdrop and the size. A real frame from your video appears behind the words, drawn by the same code that writes the file, so what you approve is what gets encoded.
Export and download. Press Add the text. Each frame is decoded, painted on and encoded again as H.264 in an MP4, and the soundtrack comes across with it: kept packet for packet when it is already the AAC an MP4 wants, encoded again when it is not, as the Opus in a WebM has to be. The line under the finished file names which of the two your clip got. Watch the bar, cancel if you change your mind, and download the result when it finishes.
What you get
- Drawn at the video's own resolution
- The words are laid out in the frame's real pixels, and the preview is that same drawing scaled down to fit your screen — never a separate approximation that looks right in the browser and wrong in the file. A source with an odd width or height is encoded one pixel narrower or shorter, because H.264 needs even sides: a 641 × 361 screen capture comes back as 640 × 360.
- The typeface is loaded before the first frame
- A canvas does not wait for a web font: text painted a moment too early comes out in the system font with no error and no warning, and you would only notice in the finished video. The faces are fetched for the characters you typed and awaited before anything is encoded, so the first second and the last one carry the same letters.
- Readable over any frame, on purpose
- White letters on a snowy shot vanish, and you cannot check every frame before exporting. So the outline and the box always take the opposite ink of the letters: black behind white and yellow, white behind black. Turning both off is one click away when your footage is dark and even, and the tool says what you are giving up.
- Nothing is added but your words
- No badge in the corner, no bar across the bottom, no ceiling on the length or the resolution, and no account to make first. Veed stamps its free exports and caps them at 720p and ten minutes; Kapwing stamps them too and stops the export at one minute. The only trace of us is the file name, which you are free to change.
Questions people ask
How do I add text to a video for free without a watermark?
Open this page, drop your video on the box, type what you want on screen and press Add the text. The MP4 you download carries your footage and your words and nothing else: no logo, no corner badge, no length cap. It is free because the work runs in your browser rather than on a machine we rent, so there is nothing for us to meter and no reason to charge for it.
Can I put text on a video on my phone?
Yes, if the browser has WebCodecs: current Chrome on Android and Safari on iOS 26 or newer qualify, and a browser that does not is told so before anything starts instead of failing halfway. Pick the clip from your camera roll and work as you would on a laptop. Phones hold less in memory, so a very long recording is better trimmed before it is written on.
Why does my text disappear behind the TikTok buttons?
Because those apps draw their own interface over your video: the description, the sound name and the row of buttons sit across the lower part of a vertical frame, and the words underneath are simply covered. That is why the position here starts at the top for portrait clips. The safe-zone checker draws each app's overlay over a still you give it, so export one frame as a PNG or a JPEG and you can see what is covered before you write anything.
What font is used, and can I change it?
One typeface, the one this site is set in, drawn at a heavy weight so it holds up against moving footage. You cannot upload your own, and that is a deliberate limit rather than an oversight: loading a stranger's font file would mean sending it somewhere or trusting it in the page. Characters the face does not carry fall back to your system's font, so other scripts still appear.
How do I show the text only during part of the video?
Switch on “Only for part of the video”, then drag the two handles along the timeline or type a start and an end such as 0:04.5. Both ways land on a tenth of a second — a typed 0:04.55 is pulled onto the grid rather than kept to the millisecond, so the field never shows one time while the file carries another — and the only value off that grid is the end of the clip, which is wherever your footage stops. Frames inside that stretch are painted, frames outside it are left as they were, and the preview jumps to the middle of the stretch so you are judging a frame the words will really land on.
Does adding text lower the quality of the video?
Every frame is encoded again, because letters painted into a picture become part of it and no frame can be copied untouched. The resolution and the frame timing are kept exactly, and the new file follows the encoder's own quality setting, so it can come out slightly larger or smaller than the one you started with. The sound is a separate story: an MP4 whose audio is already AAC keeps those packets as they are, while audio that an MP4 cannot hold — the Opus in a WebM, the Vorbis in an MKV — is encoded again and does lose a generation. The result line says which of the two your file got. What you never lose is a generation to an upload and a download.
Is my video uploaded to put the text on it?
It is not. The page reads the file straight from your disk and builds the new one in the tab's own memory; the only thing fetched while you work is the tool's code and the typeface. Nothing about the footage travels, which is the part that matters when the clip is an unreleased edit, a client's material or a recording of people who did not agree to be uploaded anywhere.
One line is typing. A whole video is transcription.
Here you write the words yourself, once. GPT-Video listens to the speech in your footage, times every word to the frame it is said on and animates the captions through the clip, then reframes the shot on whoever is talking.
Related
Your file is processed on your device and never uploaded.