Thanks

Everyone who sent something in

The archive only grows because people find a picture that is missing and take the trouble to send it in. Every one of them made this bigger than one person could.

Submissions arrive without a name attached, so there is nobody to list here by name - but if you have ever used the form, this part is about you.

The AI that does the heavy work

All of it runs on this server, or on a visitor’s own PC through the worker. There is no cloud service and no API key anywhere in Shig.ee.

• Real-ESRGAN

  Enlarges pictures, and redraws them at the same size for the finish step. The ncnn-vulkan build, v0.2.5.0.

• IS-NetApache-2.0

  Cuts the subject out of the background - one model for drawings and emoji, one for photos. Distributed with rembg.

• stable-diffusion.cpp

  Runs the painter on the machine itself, with no Python and no service behind it.

• Stable Diffusion 1.5 InpaintingCreativeML OpenRAIL-M

  Paints new surroundings into empty canvas space. Quantised to GGUF by gpustack.

Built with

• Next.js 16MIT

  The framework the whole site runs on.

• React 19MIT

  Everything you click on.

• sharpApache-2.0

  Every resize, crop, conversion and preview, and the frame-by-frame work on animations.

• better-sqlite3MIT

  The library's database, a single file on disk.

• ONNX RuntimeMIT

  Runs the cut-out models.

• FFmpeg

  Reads and writes the clips.

• Fraunces & DM Sans

  The type: Fraunces for the headings, DM Sans for everything else.