Captions Overlay v1.0.0

Good captions are mostly invisible and sometimes spectacular. Captions Overlay gives your agent one simple model for every phrase: drop the filler, put ordinary words on a clean word-by-word rail at the bottom, and save the big moment for one word composited behind your subject. It also fixes the most common layout mistake: shifting everything up to make room for captions.

How it works

1. Sort: each phrase becomes drop (ums, restarts), rail (the default) or embed (the peak).
2. Rail: rail words appear one by one in a small line near the bottom, over the picture. A punch word can get a colour accent.
3. Embed: at most one big word per beat goes behind the subject, with its own entrance and exit, never two at once.
4. Layout: the composition stays centred on the true middle of the frame; nothing moves up to make room.
5. Check: frames are reviewed by eye for caption legibility, not by a fixed keep-out zone.

Tips

• Load it together with your main HyperFrames Skill; it adds rules on top of it.
• Use embeds sparingly: one per short clip is usually enough.
• For the embed you need the subject on its own layer (a cut-out or a matte).
• Keep small critical text such as a URL out of the bottom-centre strip where the rail sits.

Open source

This Skill is part of the HyperFrames repository. Made by HeyGen and the HyperFrames contributors (heygen-com/hyperframes) and shared under the Apache License 2.0. The download is the Skill's instructions; HyperFrames itself runs from npm.