David OKuniev

Something I’ve wanted to solve for a while: could help videos just make themselves? Been experimenting with the new Supercut video/motion rendering engine (dropping soon). You describe what you want to demo or explain. An agent works through the real product, and out comes a finished video, with voice-over, cursor movement, camera moves and zooms. Your brand kit goes on top: title cards, step labels, outros. All controlled through prompts and saved as reusable dynamic components. I tried it two ways. One is an automated headless recording of the app. The other is fully synthetic: every button, label and panel is rebuilt as its own editable layer. You can rename things, swap in different data or restyle the whole thing, even after the video is finished, without re-recording anything. Can you tell which is which?

@supercut_video soon.

AI driven motion/video editing with a multi-canvas editor, been having fun with mixture.dev this weekend. Its great having it generate video based on the music and a simple prompt. We're also working on a big update to supercut.ai which will benefit from this engine.