How it works
Video Magic teleports you from your video into a world of your choice. Four steps, and you see what you are buying before anything is charged.
Record yourself
Phone on a tripod, vertical, main lens. Stand still and talk. You never walk, because every bit of movement in the finished video comes from the camera travelling through the world around you. The full brief takes five minutes to read and is the difference between a video you post and one you do not.
Choose your world
A listing photo, a rendering, a hotel lobby — anything wider than a phone screen. Video Magic measures both files, finds you in your video, and works out how tall you should be and where your feet land, so you read as standing on that floor rather than pasted on top of it.
Look before you buy
Six frames from across the finished shot, laid out as a grid, with sliders for where you stand, how big you are and how far the camera travels. Every change redraws the grid in about a second. Nothing is charged until you press Render.
Render
The finished 9:16 video, with your audio, ready to post. Video Magic emails you when it lands and keeps it in your account, so you can close the tab and walk away.
What Video Magic decides for you
Where your feet land on the floor line. How tall you are at that distance. How far the camera can travel before it runs out of backdrop. Which way the light falls, so your contact shadow points the right way. And how much the whole frame should breathe, so the result does not read as two stacked videos. You can overrule any of it — every recommendation is marked on its slider rather than fenced off.
Why not a generative video model
We tried. A model asked to move a camera through a room reinvents the room as it goes: it is a plausible kitchen, not your kitchen. Anywhere the picture is of somewhere real — a property, a venue, a product — that is a serious problem, and everywhere else it is simply worse. Travelling a crop across the backdrop you supplied is free, instant and exact.