You could just print the black-and-white image directly, but taping a solid printed shape to a curved pumpkin and cutting exactly along its edge is hard to do accurately by eye. The traditional technique — poking a pin through a series of dots to transfer the pattern onto the pumpkin's skin, then carving by connecting those marks — works far better on a curved surface, and it's what real carving stencils actually look like. This tool builds that same pattern automatically from a photo, rather than a hand-drawn design.
Why doesn't the stencil view change when I toggle "Invert"?
Because it isn't supposed to — inverting swaps which side of each boundary gets carved away, but the boundaries themselves don't move, and the printed stencil is just dots marking those boundary positions. Use the carved preview to decide which side you actually want to cut, then carve accordingly once the pattern's transferred.
The preview shows floating "islands" inside a cut-out area — is that a bug?
No — it's a real limitation of automatic threshold-based detection, not a mistake. A shape like an eye that's fully surrounded by a cut-away area (rather than connected to it) will look like it's floating with no support in the preview, the same way it would if actually carved without adding a bridge. Professional stencils are often manually adjusted to add small connecting bridges for exactly this reason — worth checking the preview and adjusting your photo or the threshold if a shape looks unsupported before you commit to carving it.
What makes a photo work well for this?
A clear, high-contrast subject against a plain background works best — the tool can't tell the difference between "background" and "subject," it only looks at brightness. For a busy or cluttered photo, try the Background Remover first to isolate the subject, then upload the result here.
Is my photo sent anywhere?
No. Everything — the conversion, the boundary detection, the stencil itself — happens locally in your browser. Nothing is uploaded to a server.