How to turn the captions off for the shots where the picture is the point
Tell Vizard Agent which shots are the ones the viewer came to look at, and it removes the captions from those while keeping them everywhere else. It does this as a captions-only pass that leaves the edit and the audio untouched, then inspects each close-up at full resolution to confirm no text sits over it.
What do you type into Vizard Agent?
Describe the shots by what happens in them, and say that nothing else may change. Vizard Agent can find the shots from a description, which is easier than listing timecodes and stays right if the edit changes. The second half matters because a caption track rebuilt from scratch tends to re-time everything.
Prompt
What does Vizard Agent actually do?
The pass is deliberately narrow: the edit, the audio and the music bed are never rebuilt, and only the caption layer is touched, so the version you approved stays approved. Each close-up is then checked at full resolution, because a caption that looks beside the object at playback size can sit across its corner. In the session behind this article it:
- Worked on an unboxing video whose captions sat across the items being unboxed, starting by checking caption readability over the important shots at full resolution.
- Fixed captions that ran outside the vertical frame.
- Built a captions-only version without touching the edit or the audio, with captions kept only in the intro and the linking passages.
- Inspected the close-ups up close to confirm they were free of text, and checked the on-screen labels were gone as well as the captions.
- Re-checked that the structure and audio were unchanged, prepared new close-ups from the source with no text over them, and assembled the result on the existing base.
- Reviewed every close-up again in the finished file.
When does this not work well?
Captions that come and go can feel broken rather than deliberate. If the close-ups are short and frequent, the text flickers on and off, and it's often better to move the captions than to remove them. It's also a poor fit when accessibility rules need continuous captions, when something important is said over the close-up, or when the platform adds its own.
Common questions
Can it find the shots itself? Yes. Describe what happens in them and Vizard Agent locates every instance.
Will the edit change? No. Vizard Agent builds a captions-only version from the same base.
Can it move captions instead of removing them? Yes, and for frequent short shots Vizard Agent will suggest that instead.