Loading page
About this tool
Compare two screenshots and detect visual differences with pixel-level accuracy.
The Screenshot Change Detector compares two screenshots pixel by pixel and marks every changed pixel in red, using a per-pixel threshold of |ΔR| + |ΔG| + |ΔB| greater than 30 across the three colour channels. It reports the total pixel count, how many changed, how many did not, and the change as a percentage to two decimal places, then lets you download the annotated diff as a PNG. It is built for anyone who needs to prove a visual regression happened rather than argue about whether the page 'looks different'.
Open Screenshot Change Detector on AltFTool — it loads instantly in your browser.
Upload two screenshots into the 'Before' and 'After' zones (any image/* file).
Click 'Compare' to flag every pixel where the red, green and blue differences sum above 30.
Read Total Pixels, Changed, Unchanged and Change %, then click 'Download Diff PNG' (saves screenshot-diff.png) or 'Copy Report'.
Changed pixels, unchanged pixels and a percentage to two decimals go into a copyable report you can paste into a ticket.
Unchanged pixels keep the before image, so the red mask sits in context and you can see what component moved, not just where.
The threshold of 30 summed across R, G and B ignores single-value compression wobble while still catching a one-pixel border or a shifted baseline.
A pixel counts as changed when the absolute differences of its red, green and blue values add up to more than 30. That means a change of about 10 per channel, or a single channel shifting by 31, is flagged — enough to catch antialiasing changes on text while ignoring the tiniest JPEG artefacts.
Both are drawn onto a canvas sized to the larger width and larger height of the pair, which stretches the smaller image to fit. That makes the comparison unreliable for mismatched captures, so take both screenshots at the same viewport size and device pixel ratio for a meaningful result.
Usually because something moved rather than changed appearance. A one-pixel shift in a body of text redraws every glyph, so the diff legitimately reports thousands of changed pixels — read the red mask's shape, not just the percentage, to tell a reflow from a recolour.
No — only the red, green and blue channels are differenced; the alpha channel is ignored. Two images that differ only in transparency over identical colour values will report 0.00% change, so flatten PNGs onto a known background before comparing if opacity matters.