BlazeDiff vs honeydiff

@blazediff/core-native is ~3.2x faster than honeydiff over our 34 fixtures, with image decode and diff PNG writing included. Both are Rust image diffs that ship as native Node addons. The difference is scope and openness: BlazeDiff publishes its full source and also runs in browsers, Python and Rust.

honeydiff is the diff engine of the Vizzly visual testing service, published on npm as @vizzly-testing/honeydiff. Its binaries carry an MIT license, but its public repository holds only a README and a changelog, not the Rust source. It returns rich JSON: changed-region statistics, perceptual scores, and accessibility checks.

At a glance

Benchmarks

Both libraries run at their defaults with anti-aliasing on and write a diff PNG. honeydiff thresholds on CIEDE2000 and BlazeDiff on YIQ, so the chart compares speed, not identical output.

Switching

honeydiff returns different and a pixels.changed.total count. BlazeDiff returns match, a reason, and a diffCount.

// Before
import { compare } from "@vizzly-testing/honeydiff";
 
const before = await compare("baseline.png", "current.png", {});
if (before.different) console.log(`${before.pixels.changed.total} pixels differ`);

// After
import { compare } from "@blazediff/core-native";
 
const after = await compare("baseline.png", "current.png", "diff.png", {
  antialiasing: true,
});
if (!after.match && after.reason === "pixel-diff") {
  console.log(`${after.diffCount} pixels differ`);
}

For the region data honeydiff puts in its result, run difference analysis. It labels each changed region as an addition, a deletion, a shift, a recolor or rendering noise.

When to keep honeydiff

• You use Vizzly, which builds on it.
• You want region statistics, perceptual scores and contrast checks from one call.
• You need CIEDE2000 color thresholds, or alignment of inserted rows.

Related

• BlazeDiff vs odiff
• Pixel comparison in native Node
• Image difference analysis
• All benchmarks