Smart guides

Illustrator called them smart guides, InDesign copied the name, and Figma, Sketch and every canvas tool since have shipped the same idea with the same red or magenta hairlines: lines that are not in the document at all. A ruler guide is something you place and it stays. A smart guide exists for as long as your hand is moving an object and vanishes on release, because it is not a guide in the drawing sense. It is feedback, drawn to explain a snap the tool has already decided to perform.

What the lines report is a small set of coincidences: the moving object’s left, centre or right edge sitting on a neighbour’s left, centre or right, the same three along the other axis, and in the better implementations equal spacing between three or more objects, which shows up as a pair of matched distance badges rather than a line. The badges are the second half of the vocabulary. A line says these two things agree; a number beside it says how far apart they are, which is what turns a nudge-until-it-looks-right session into a measured one. Anything the tool can measure it can also offer to match, so the same overlay that reports a 36 pixel gap will happily snap the next object to 36.

The craft is almost entirely in the snap radius and in what the tool refuses to say. Too small a radius and the guides feel unreliable, since the object has to be nearly aligned before the tool admits it. Too large and the object is yanked off positions the person actually wanted, which is why every tool with snapping also has a modifier key that suspends it for the length of one drag. The radius should also be measured in screen pixels rather than document units, so it stays the same size under the hand at every zoom level. And the display has to stay quiet: showing every coincidence at once produces a plaid of lines that reports nothing, so tools rank candidates and draw the nearest one or two.

Grid snapping is the useful contrast, and the site has snapping for that broader sense. A grid is constant, invisible and indifferent to content: everything lands on a multiple of eight whether or not anything else is there. Smart guides are the opposite on every count. They are derived from whatever happens to be nearby, they only exist during the gesture, and they are the reason a person can align to another object without ever creating a guide, opening an alignment panel, or typing a coordinate. Neither replaces the other, which is why serious tools run both and let the nearer one win.