- What kind of photos work best for silhouette conversion?
- Portraits and photos where the subject is clearly distinguishable from the background give the best results. Front-lit subjects against a plain or blurred background are ideal. Profile shots (side-on) are particularly striking as silhouettes because the outline of a face is instantly recognisable. The AI works on any subject — people, pets, objects — but portrait photos tend to produce the most dramatic backlit art effect.
- Is my photo uploaded to a server?
- No. The entire process runs inside your browser using WebGL via TensorFlow.js and MediaPipe. The segmentation model is downloaded once (a few MB, cached for repeat use) and then runs locally on your device using your GPU. Your photo data never leaves your machine — there are no analytics or server logs capturing your images.
- Why does it take a moment the first time I use it?
- The first time, your browser downloads the TensorFlow.js and MediaPipe model files. These are cached by your browser, so subsequent conversions on the same device are much faster — typically just a few seconds. Processing time after that depends on your photo size and device speed; most portraits complete in 2–8 seconds.
- Can I use the silhouette art commercially or print it?
- The tool produces images from your own uploaded photos. You own the output. The background gradients are generated entirely by canvas code — no third-party images are used. As long as you own or have permission to use the original photo, the resulting silhouette art is yours to use as you wish — print, share, or use in personal and commercial projects.