Apple Vision builds a mask
HookWise runs the clip through Apple Vision, the person segmentation built into macOS. For each frame it builds a grey mask, white where a person is and black everywhere else. The mask is made at 15 frames per second and 512 pixels wide, then scaled to your clip during preview and render.
The work runs as a background job on your Mac, and the video never leaves your computer. Once the mask is ready, whatever sits on the track below shows through behind the person: B-roll, a product video or another recording. Position and scale it with keyframes like any other layer.
Recording tips for a cleaner cut
Help the mask by making the person stand out from the room. Light your face evenly, avoid a background that matches the color of your shirt, and leave some space between you and the wall. For a first try, pick a clip where the speaker sits still, such as a talking-head tip or a product review at a desk.
The mask updates 15 times per second, so fast hand movement and loose hair are the moments to check in the preview. Place the new background on the track underneath, then use keyframes to move or scale it slowly so it does not look pasted on.
Requirements and the Windows option
Background removal without a green screen needs macOS 12 or later. It runs offline, uses no Gemini key and is not one of the AI plan features. It is not available on Windows yet. Windows users with a green screen can use chroma key, which has a despill option to remove the green tint from edges of skin and hair.
Processing time depends on the clip length and your Mac, and HookWise lists no RAM or GPU requirement. Try a short clip first to see how long it takes on your machine. Run it after trimming, so HookWise only processes the parts you keep.
