Convert an image into pixel art
Upload an image, choose a canvas size and refine the result. Export a PNG when you are ready.
Open the editor ↗A face can be recognizable from a single edge. Explore twelve pixel-art adaptations of William Bache’s historic silhouette portraits: dark profiles, pale paper and small changes of forehead, nose and chin.
Inspired by the Smithsonian Sidedoor episode The Toxic Book of Faces, this collection turns digitized museum objects into 128×128 studies you can use as a starting point for your own portrait ideas.
Each example comes from a different CC0 collection image in Bache’s silhouette ledger. The gallery contains eleven PNGs and one whole-image floating GIF; the original sitters have not been redrawn as fictional characters.












For inspiration only. Rights remain with their owners; source licenses apply. Contact us for removal requests.
Three ways to start. One pixel editor.
Upload an image, choose a canvas size and refine the result. Export a PNG when you are ready.
Open the editor ↗Add or duplicate frames, adjust the timing and preview the loop. Export your animation as a GIF.
Open the editor ↗Start with a blank canvas. Use the drawing and fill tools, then export your icon with a transparent background.
Open the editor ↗See how creators use the app and make their displays their own.
More ideas, more tools and a place to share.
Browse community artwork and save the designs that catch your eye.
Draw, edit and animate in the Divoom app.

See what other creators are making and share your own work.
Divoom app reviews from Google Play.
It refers to arsenic found in the physical Bache album, as described by the Smithsonian’s Sidedoor episode and the National Portrait Gallery. It does not refer to a style of face, a fictional creature or a green color effect.
They are pixel-art adaptations of twelve separate William Bache silhouette images. Some sitters are named in the museum records; others remain unidentified. Each source record and object number is retained in the sources below.
The twelve media files selected for this page are explicitly designated CC0 in Smithsonian Open Access records. That designation applies to these selected collection images; it does not automatically cover podcast cover art or every image on si.edu. Follow the linked object records and Smithsonian usage terms for the material you choose.
Preserve the biggest changes in the side profile first, especially the nose projection, chin angle and the gap below the lower lip. Use a 32×32 or 64×64 working grid and leave breathing room in front of the face. Fine paper texture matters less than a clear outline.
The final portrait shifts as one complete image in a short floating loop. Its mouth, eyes and head do not move independently; it is an animated presentation of an adaptation, not a reconstruction of the historical sitter.
Try a side-profile photo of your own, simplify its outer edge and choose a small palette. Use the historic silhouettes above as a lesson in contour rather than copying their paper texture pixel for pixel.