Video editing guide

Add Text to Video Online with Exact Timing

A useful video title is readable, correctly spelled and visible for the right amount of time. Add a timed text layer manually or ask the AI agent for the exact words and placement, then review it over the moving footage.

Choose what the text needs to communicate

An intro title, a short instruction and a label for an object need different durations. Keep the wording brief enough to read while the video continues. A long sentence at the bottom of a fast demonstration makes viewers choose between reading and watching the action.

Decide where the layer belongs before adding it. The bottom may be appropriate for a title, but it can also cover subtitles already burned into the source, a product label or a player control. Check the whole interval where the text will appear rather than choosing the position from one frame.

Add a text layer with a start and end

Move to the intended point on the current timeline and use Add text, or describe the layer in an AI request. Enter the exact words and choose the top, center or bottom position. Set a start and an end that fit within the video’s current duration.

Times belong to the current project, not necessarily the original file. If you removed an opening section, the intro in the edited version begins at a new source position. Use the time shown on the project timeline when asking the agent to add or change a layer.

Add “Welcome to the demo” at the bottom from 00:00 to 00:03. Use white text with a dark background.

Set contrast and size for the finished frame

White text with a dark background is a practical starting point over changing footage. If you use another color, test it against both light and dark moments within the layer’s duration. A title should remain readable when the background changes rather than only at its first frame.

Review at the intended output shape and viewing size. A text layer that looks comfortable on a desktop preview can be too small on a phone. Avoid filling the frame with a paragraph; split a long instruction into shorter timed layers where that helps the viewer follow the action.

Check when text appears and disappears

Play from just before the layer starts to just after it ends. The words should arrive when they become relevant and stay long enough to read. If the text is missing in a still preview, confirm that the playhead is inside the layer’s time range before assuming it failed.

When you cut or rearrange the video, recheck the text timing. A title may no longer describe the material underneath it. Edit the layer itself rather than adding another overlapping title, especially if the same words would appear twice in the export.

Understand the difference between a title and captions

A timed text layer can show a title, label or a manually entered line of speech. It is not an automatically generated transcript. This beta does not automatically transcribe audio or create a complete set of captions from a recording.

If you enter spoken lines yourself, verify them against the audio and give each line its own readable range. Do not describe a manually added intro title as automatic subtitles. A transcript-dependent request needs evidence the current agent does not receive through sparse image samples alone.

Export the video with the text rendered into it

Text is rendered into the exported picture, so viewers do not need a separate subtitle file to see it. Choose a supported MP4 or WebM export, then inspect the downloaded video at the title’s first frame, middle and last frame. Read every character before publishing.

Keep the original source and a project backup if you may need to change the words later. The exported video contains the visible text in its picture; the project contains the editable layer. Browser storage can retain the project, but clearing it can remove local media and edit records.

Frequently asked questions

Can I add text only to the first three seconds?

Yes. Set the text layer to start at 00:00 and end at 00:03, or include those times in your prompt.

Will the text appear in the exported MP4?

Yes. Timed text is rendered into the video frames during export. Check the downloaded file to verify its wording, timing and readability.

Does this automatically generate subtitles?

No. Automatic transcription and captions are not included in the current beta. You can manually create timed text layers.