---
name: Smart guides
slug: smart-guides
category: interaction
status: published
created: 2026-08-21T00:00:00.000Z
modified: 2026-08-21T00:00:00.000Z
definition: Alignment lines that appear only while dragging, showing which edges
  or centers the moving object currently lines up with.
aliases:
  - name: alignment guides
    source: community
  - name: dynamic guides
    source: community
  - name: magnetic guides
    source: community
  - name: red lines
    source: community
tags:
  - canvas
  - design-tools
relations:
  contrastWith:
    - snapping
  variantOf: []
  partOf: []
  seeAlso:
    - keyline
implementations: []
sources:
  - title: "CreativePro: Smart(er) guides"
    url: https://creativepro.com/smarter-guides/
demo: inline
exhibit: false
useWhen: the guide lines that appear mid-drag
---

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.
