Package net.fabricmc.fabric.api.client.gametest.v1.screenshot

Interface TestScreenshotComparisonAlgorithm

public interface TestScreenshotComparisonAlgorithm

An algorithm for finding a template subimage (the "needle") in a screenshot (the "haystack"). Comparison algorithms are written to find a subimage, but can also be used to compare two images of equal size (which is a special case of finding). Custom algorithm implementations are allowed.

Nested Class Summary

Nested Classes

Modifier and Type	Interface	Description
static interface	TestScreenshotComparisonAlgorithm.RawImage<DATA>	A thin wrapper around a raw image data array, so that algorithms can directly access the array which may be more efficient than going through a NativeImage each time.

Method Summary

Modifier and Type	Method	Description
static TestScreenshotComparisonAlgorithm	defaultAlgorithm()	The default algorithm for this API, which is the Mean Squared Difference algorithm with a threshold of 0.005.
static TestScreenshotComparisonAlgorithm	exact()	An algorithm which searches for an exact match of the needle in the haystack.
@Nullable org.joml.Vector2i	findColor(TestScreenshotComparisonAlgorithm.RawImage<int[]> haystack, TestScreenshotComparisonAlgorithm.RawImage<int[]> needle)	Finds a needle in a haystack, color image version.
default @Nullable org.joml.Vector2i	findGrayscale(TestScreenshotComparisonAlgorithm.RawImage<byte[]> haystack, TestScreenshotComparisonAlgorithm.RawImage<byte[]> needle)	Finds a needle in a haystack, grayscale image version.
static TestScreenshotComparisonAlgorithm	meanSquaredDifference(float maxMeanSquaredDifference)	The Mean Squared Difference algorithm computes the mean of differences between the needle pixels and corresponding haystack pixels squared, and checks if the mean is less than a particular threshold.

Method Details

defaultAlgorithm

static TestScreenshotComparisonAlgorithm defaultAlgorithm()

The default algorithm for this API, which is the Mean Squared Difference algorithm with a threshold of 0.005. See meanSquaredDifference(float) for details.

Returns:

The default algorithm for this API

meanSquaredDifference

static TestScreenshotComparisonAlgorithm meanSquaredDifference(float maxMeanSquaredDifference)

The Mean Squared Difference algorithm computes the mean of differences between the needle pixels and corresponding haystack pixels squared, and checks if the mean is less than a particular threshold.

Parameters:

maxMeanSquaredDifference - The maximum mean squared difference between the needle and the subimage in the haystack for a match

Returns:

The Mean Squared Difference algorithm with the given threshold

exact

static TestScreenshotComparisonAlgorithm exact()

An algorithm which searches for an exact match of the needle in the haystack. This is simpler and faster than a fuzzy match, but is prone to failing on even the slightest inconsistency in the image.

Returns:

An algorithm for exact matches

findColor

@Nullable org.joml.Vector2i findColor(TestScreenshotComparisonAlgorithm.RawImage<int[]> haystack,
 TestScreenshotComparisonAlgorithm.RawImage<int[]> needle)

Finds a needle in a haystack, color image version. Pixels are in RGB format (with no alpha channel).

Parameters:

haystack - The image in which to search for the needle

needle - The image to search for

Returns:

The location of the match, or null if no match was found. In case of multiple matches, returns an arbitrary match

findGrayscale

default @Nullable org.joml.Vector2i findGrayscale(TestScreenshotComparisonAlgorithm.RawImage<byte[]> haystack,
 TestScreenshotComparisonAlgorithm.RawImage<byte[]> needle)

Finds a needle in a haystack, grayscale image version. Each pixel is a brightness value from 0-255.

Note for custom algorithm implementations: the default implementation converts the images to color and then uses findColor(net.fabricmc.fabric.api.client.gametest.v1.screenshot.TestScreenshotComparisonAlgorithm.RawImage<int[]>, net.fabricmc.fabric.api.client.gametest.v1.screenshot.TestScreenshotComparisonAlgorithm.RawImage<int[]>), but it is usually possible to implement this more efficiently. You should strongly consider overriding this method.

Parameters:

haystack - The image in which to search for the needle

needle - The image to search for

Returns:

The location of the match, or null if no match was found. In case of multiple matches, returns an arbitrary match