Can I compare images of different sizes?

Yes. Both are placed at the top-left of a canvas sized to the larger width and the larger height, and the area where the smaller image doesn't reach is filled white — which then shows up as a difference. That's usually what you want (a size change is a change), but if you only care about content, crop or resize the images to matching dimensions first.

How sensitive is the comparison?

A pixel counts as different when any channel differs by more than 30 out of 255. That threshold is fixed — it's deliberately tolerant enough to ignore JPEG re-encoding noise while still flagging any change you could plausibly see. For bit-exact verification needs, compare lossless PNGs, where any genuine difference will comfortably exceed the threshold.

Why does a tiny layout shift light up huge areas in red?

The comparison is positional: pixel (x, y) in one image against pixel (x, y) in the other. If content moved by even one pixel, every edge in the moved region lands on different coordinates and gets flagged. A large red percentage therefore means 'much of the image is in a different place', not necessarily 'much of the content changed'.

What does the percentage actually measure?

The share of all canvas pixels that exceeded the threshold, to two decimal places. 0% means visually identical within tolerance; fractions of a percent usually mean small, localized edits.

Diff Images

Compare two images pixel by pixel. Differences are highlighted in red.

Drop image files here or browse

Max 50MB per file · Up to 2 files

About Diff Images

Comparing two images by eye is unreliable — a one-shade color change or a 2-pixel layout shift is invisible until a customer notices it. Convertora's diff tool compares two images pixel by pixel and renders a result where every changed pixel is painted solid red over a dimmed copy of the original, along with the exact percentage of pixels that differ. What changed, and how much, stops being a judgment call.

The comparison runs on a canvas in your browser: both images are decoded locally, their pixel data is compared channel by channel, and the diff image is assembled without anything being uploaded. A small built-in tolerance (a difference of 30 per color channel) keeps ordinary JPEG compression noise from lighting up the whole image, so the red areas are real changes, not encoding jitter.

How to use it

1Upload exactly two images — JPG, PNG, WebP, or AVIF, up to 50 MB each. The file names being compared are shown so you know which is which.
2Click Compare. Both images are drawn onto a shared canvas sized to the larger of the two in each dimension, with any uncovered area filled white.
3Each pixel pair is compared channel by channel. Pixels differing by more than 30 in red, green, or blue are marked; smaller variations are treated as identical.
4View the result: changed pixels in solid red, unchanged content dimmed underneath for orientation, and a stat line with the percentage of differing pixels.
5Download the diff as a PNG to attach to a bug report or review thread.

Common use cases

Catching visual regressions by comparing before/after screenshots of a UI change.
Verifying that a re-export or format conversion didn't subtly alter a graphic.
Spotting the difference between two near-identical product photos or document scans.
Confirming that two renders from different machines or browsers are actually identical.

Frequently asked questions

Tips

Compare PNG screenshots rather than JPG ones when you can — lossless sources mean every red pixel is a real change rather than a compression artifact that crossed the threshold.
If you only care about one region, crop both images to it first (Crop Image tool) — the percentage then measures the area you care about instead of being diluted by unchanged surroundings.

100% private — runs in your browser

Convertora processes everything on your device using JavaScript and WebAssembly. Files never leave your browser, are never uploaded to a server, and are never seen by us or anyone else. The moment you close the tab, the data is gone — there is no temporary cloud copy, no log entry, no retained backup.

Because the work happens locally, processing speed depends on your device — but there are no rate limits, no daily caps, and no file size restrictions beyond what your browser can handle in memory. No signup, no account, no payment. The tool works the same in incognito mode, on a corporate network, or after the page has loaded once, even with the network disconnected.