About Background Studio

Recording a short talking-head video — an interview answer, an introduction, an explanation of something — should take about as long as the video. In practice the options were: record your real room and accept it, upload your footage to a service that would create an account for you first and possibly watermark the result, or install a desktop application to get a virtual background and then find something else to record with.

None of those is unreasonable. They are just all disproportionate to the task. The annoying part was that the browser already had every piece needed: it can open a camera, run a segmentation model, composite frames on a GPU and encode a video file. The pieces were sitting there unassembled.

No account, because there is no server. Not “we do not require one” — there is no backend in the project at all, so there is nothing to have an account with. That single decision removes sign-up, password resets, data retention, and the question of what happens to your video if the service shuts down.

The recorder is fed the same canvas you are watching. There is exactly one rendering path, so the file cannot come out looking different from the preview, and there is no export step in which something could go wrong. It also means the raw camera is never handed to the recorder, so an unprocessed frame has no route into the output.

The privacy claim is enforced rather than promised. A Content-Security-Policy tells your browser to refuse connections to any other host, and the machine-learning model is served from this domain so that policy can stay that tight. That is a deliberately stronger arrangement than a written assurance, because it does not depend on the code — or its dependencies — behaving. The details, including what it does not claim.

Nothing is faked. If a browser cannot record, the studio says which capability is missing rather than showing a button that does nothing. If segmentation fails to load, it passes the camera through unmodified and tells you. There is no placeholder anywhere that looks like it works.

It is not a video editor: no timeline, no trimming, no titles. It is not a screen recorder. It is not a virtual camera, and cannot be — a browser tab has no way to present itself to your operating system as a webcam, so it will never feed a processed image into Zoom or OBS.

And it is not trying to become any of those. The value is in doing one thing without ceremony; each of those additions would bring an account, a server, or an install back with it.

One developer, as a working answer to their own problem rather than as a product with a roadmap. There is no company behind it, no funding to recover and no paid tier planned — which is worth saying plainly, because “free” usually means something is being collected instead, and here nothing is.