How to mirror a video without reversing the text on screen
Upload the video and tell Vizard Agent to mirror everything except the words. Vizard Agent finds where every caption and sticker appears, isolates them onto a transparent layer at full resolution, flips the footage underneath, and lays the text back over it the right way round.
What do you type into Vizard Agent?
Say what to flip and what to exempt in one sentence. Vizard Agent treats the exemption as the harder half of the job, since the text has to be found and lifted off, and stating it up front is what makes the approach different from a one-line flip. Mention any logo or symbol that must also stay readable.
Prompt
What does Vizard Agent actually do?
Flipping a video horizontally is a single command. Flipping it without turning every caption into mirror writing is a different job: the text has to be lifted off first, the picture flipped underneath, and the text laid back on top using the original pixels, so nothing needs retyping. In the session behind this article it:
- Looked at real frames and the video's parameters, and checked the shot list to see where the captions sat and how they moved, before flipping anything.
- Reviewed an extended storyboard of the whole clip and examined the stickers and text close up, including the opening sticker, the subscribe caption and the question mark.
- Determined the exact boundaries where each text panel appeared, since a boundary a few pixels off leaves a visible seam beside the restored caption.
- Isolated the text regions at source resolution and verified each crop by eye, twice.
- Built a transparent layer with only the text characters, mirrored the footage and composited the text back the right way round, delivering 52.44 seconds at 360x640.
When does this not work well?
Lifting text off a video means reconstructing whatever sat behind it, and that gets harder as the text or the background moves. Travelling captions need tracking, semi-transparent ones blend with the background, and text over faces is awkward. Some things should flip, like a sign in the scene, and mirroring reverses handedness and road direction too.
Common questions
Will the captions be retyped? No. Vizard Agent reuses the original pixels, so the typeface and styling are exactly right.
What about text on a sign in the scene? That usually should flip. Say so and Vizard Agent leaves it in the mirrored layer.
Why lift the text off instead of flipping it back in place? Because flipping a region twice leaves the background inside it mirrored the wrong way. Vizard Agent keeps the picture and the words on separate layers.