An image may be displayed smaller than its decoded pixel grid. PixelPick keeps the Canvas at decoded width and height, measures its rendered rectangle, scales the pointer offset back into source space, floors each coordinate and clamps the outer edge. The displayed x and y start at zero, so the upper-left pixel is 0,0 and the lower-right is width minus one, height minus one.
The magnifier does not invent pixels
An 11×11 neighborhood is copied into a small preview with image smoothing disabled. Each source pixel becomes a crisp block and the center marker identifies the value that a click will lock. Near an image edge the neighborhood is clipped rather than wrapped. Arrow keys move the locked point one source pixel at a time, which helps when touch or a scaled display makes a one-pixel target difficult.
Sampling happens after browser decode
JPEG compression, WebP decoding, transparency and an embedded color profile are resolved before Canvas exposes RGBA bytes. That makes the result useful for the browser rendering workflow but not a claim about raw file bytes or another color-managed application. PixelPick rejects animations so the sampled frame cannot change silently while a coordinate stays the same.
Know what the value describes
PixelPick reports the RGBA bytes produced after this browser decodes the file into Canvas. Embedded profiles, browser conversion and display characteristics can make those bytes differ from another workflow. Use the result for ordinary digital design inspection, then verify important brand or production colors in their authoritative source. It is not calibrated colorimetry, physical paint matching, medical interpretation, accessibility certification or a safety measurement.