Where Confetti Beads Come From: Dithering vs. JPEG Compression vs. Over-Paletting
Understanding why raw converters litter portraits with confetti beads reveals why simply dragging a "Max Colors" slider down to 10 rarely fixes a messy chart. Three distinct technical flaws generate stray single-bead islands during photo conversion:
First, Floyd–Steinberg or Atkinson error-diffusion dithering intentionally pushes quantization error onto neighboring right and bottom cells, sprinkling alternating tan, grey, and pink pegs across smooth skin. Dithering works on 300-DPI paper printers; on a 5-DPI pegboard (5.00mm beads), it looks like chickenpox. Second, unfiltered backgrounds and JPEG edge ringing leak green lawn and couch-fabric hues into fur silhouettes. Third, unconstrained 100-color palettes assign five nearly identical warm greys (ΔE00 < 2.0) to a single shadow, splitting one shadow zone into fragmented micro-islands.
| Pipeline Configuration | Active Bead Shades | Isolated 1–2 Peg Islands | Estimated Crafting Time |
|---|---|---|---|
| Legacy Weighted RGB + Dithering | 34 colors (incl. clear/glow) | 410+ stray confetti pegs | ~3.2 hours (constant tray swaps) |
| Area Downsample + CIEDE2000 Solid Lock | 16 solid opaque colors | ~95 minor edge specks | ~2.1 hours |
| + Free Auto De-Confetti (|ΔL| ≥ 26 Lock) | 14 solid opaque colors | < 12 intentional highlights | ~1.5 hours (clean color blocks) |
| + Optional Flat AI Pre-Chart (Cluttered BG) | 12–15 crisp sprite zones | Zero background noise | ~1.3 hours |