Use case · visual regression

A diff that means the page changed.

Byte-identical output for the same request against the same renderer version, so a difference between two runs is a difference in the page rather than in the machine that photographed it.

Start free See the parameters

0 px changed

visual regression Visual regression Two runs, same bytes. When the diff lights up, the page really changed.

the problem

Most of the noise is not the page.

Most visual regression noise is not the page. It is a font that loaded differently, an emoji from another platform's sheet, an animation caught mid-frame, or a scrollbar that was there once. You end up either ignoring the diff or maintaining a tolerance nobody trusts.

the answer

Pin it, clean it, wait for it.

Pin the renderer version, block the ad networks, hide the carousel, and wait properly — the load event, network idle, every image decoded, the fonts painted, then any delay you ask for. What is left over is the page.

// the call
{
  "url": "https://example.com",
  "full_page": true,
  "block": ["ads","trackers","consent"],
  "delay_ms": 500
}

0 px changed

Every response carries the renderer version it was taken with, so a diff can tell you whether the browser moved underneath you.