Skip to content

Visual Testing Settings

Visual testing settings control how Vizzly finds image changes, exposes screenshots, and adds images to GitHub comments. Start with the defaults. Change comparison settings only when your screenshots need different sensitivity or noise filtering.

Comparison sensitivity

Vizzly uses a CIEDE2000 Delta E threshold for color differences.

Common values:

  • 1.0 catches subtle changes
  • 2.0 is the default and the best starting point
  • 3.0 to 5.0 ignores more small rendering differences

Changes below the threshold are treated as identical.

Noise filtering

Minimum cluster size controls how many connected pixels must change before Vizzly counts the area as a diff.

  • 1 counts every different pixel
  • 2 is the default and removes isolated single-pixel noise
  • 5 or higher keeps only larger connected areas

Higher values reduce noise, but they can hide very small changes. Raise this setting carefully.

Diff highlight color

Choose the color used to mark changed pixels in generated diff images. Pick a color that stands out from the UI you test most often.

Baseline signature properties

Vizzly always uses the screenshot name, viewport width, and browser to match a screenshot with its baseline. Add signature properties when one name needs separate baselines for different variants.

Common examples include theme, device, and locale.

See Baselines for the full matching and selection behavior.

Public properties

Public properties expose matching screenshots without normal authentication. Each rule matches one metadata key and value.

Use them carefully. They change who can view a screenshot.

GitHub comment images

When PR comments are enabled, choose which images Vizzly includes:

  • no images
  • new screenshots
  • diff images

Next steps