ShapeAddress 0.3.0 · lab
Real pixels, every dictionary, every layout
Pick a dictionary and a layout. The page draws the address at 16 to 64 px on light and on dark, enlarges the raster next to it, measures what each glyph gets in pixels, and drops the result into a crowd of near-identical addresses. Nothing here is scaled with CSS: the small picture is the real one.
At this size
Left: the browser's own rendering at 32 px. Right: the same picture rasterised once at 1x and enlarged with no smoothing, so each pixel is one block. On a high-density screen the left one is sharper than the right one; the right one is what a 1x screen shows.
Thresholds
Real size, 16 to 64 px
Each size twice, on white and on dark, at its exact pixel size. Next to 24, 32 and 48 px the twin shows the 1x raster at 6x.
The crowd
Twelve addresses derived from the input: one bit, one byte, two, four, eight and all twenty bytes changed, two of each. The set is seeded by the input, so it is stable. Each at 24 and 32 px on white and on dark. Your input is outlined. Shuffle hides the labels; tap the one you think is the original.
What to judge. Which layout keeps the address recognisable at 24 px. Which dictionary's glyphs survive at 6 px. Whether one changed byte shows in the crowd. The measured answer lives in the small-size study; this page is where you look at the pixels.