Introducing LiteRT: Google's high-performance runtime for on-device AI, formerly known as TensorFlow Lite. Learn more

ImageGenerator

public final class ImageGenerator

Performs image generation from a text prompt.

Nested Classes

class	ImageGenerator.ConditionOptions		Options for setting up the conditions types and the plugin models
class	ImageGenerator.ImageGeneratorOptions		Options for setting up an `ImageGenerator`.

Public Methods

void	close() Closes and cleans up the task runners.
MPImage	createConditionImage(MPImage sourceConditionImage, ImageGenerator.ConditionOptions.ConditionType conditionType) Create the condition image of specified condition type from the source image.
static ImageGenerator	createFromOptions(Context context, ImageGenerator.ImageGeneratorOptions generatorOptions) Creates an `ImageGenerator` instance from an `ImageGenerator.ImageGeneratorOptions`.
static ImageGenerator	createFromOptions(Context context, ImageGenerator.ImageGeneratorOptions generatorOptions, ImageGenerator.ConditionOptions conditionOptions) Creates an `ImageGenerator` instance, from `ImageGenerator.ImageGeneratorOptions` and `ImageGenerator.ConditionOptions`, if plugin models are used to generate an image based on the condition image.
ImageGeneratorResult	execute(boolean showResult) Executes one iteration of image generation.
ImageGeneratorResult	generate(String prompt, MPImage sourceConditionImage, ImageGenerator.ConditionOptions.ConditionType conditionType, int iterations, int seed) Generates an image based on the source image for iterations and the given random seed.
ImageGeneratorResult	generate(String prompt, int iterations, int seed) Generates an image for iterations and the given random seed.
void	setInputs(String prompt, int iterations, int seed) Sets the inputs of the ImageGenerator.
void	setInputs(String prompt, MPImage sourceConditionImage, ImageGenerator.ConditionOptions.ConditionType conditionType, int iterations, int seed) Sets the inputs of the ImageGenerator.

Inherited Methods

From class com.google.mediapipe.tasks.vision.core.BaseVisionTaskApi

void	close() Closes and cleans up the MediaPipe vision task.
static long	generateResultTimestampMs(RunningMode runningMode, Packet packet) Generates the timestamp of a vision task result object from the vision task running mode and the output packet.

From class java.lang.Object

boolean	equals(Object arg0)
final Class<?>	getClass()
int	hashCode()
final void	notify()
final void	notifyAll()
String	toString()
final void	wait(long arg0, int arg1)
final void	wait(long arg0)
final void	wait()

From interface java.lang.AutoCloseable

abstract void

close()

Public Methods

public void close ()

Closes and cleans up the task runners.

public MPImage createConditionImage (MPImage sourceConditionImage, ImageGenerator.ConditionOptions.ConditionType conditionType)

Create the condition image of specified condition type from the source image. Currently support face landmarks, depth image and edge image as the condition image.

Parameters

sourceConditionImage	The source image used to create the condition image.
conditionType	The `ImageGenerator.ConditionOptions.ConditionType` specifying the type of condition image.

public static ImageGenerator createFromOptions (Context context, ImageGenerator.ImageGeneratorOptions generatorOptions)

Creates an ImageGenerator instance from an ImageGenerator.ImageGeneratorOptions.

Parameters

context	an Android `ERROR(/Context)`.
generatorOptions	an `ImageGenerator.ImageGeneratorOptions` instance.

Throws

	if there is an error during `ImageGenerator` creation.

public static ImageGenerator createFromOptions (Context context, ImageGenerator.ImageGeneratorOptions generatorOptions, ImageGenerator.ConditionOptions conditionOptions)

Creates an ImageGenerator instance, from ImageGenerator.ImageGeneratorOptions and ImageGenerator.ConditionOptions, if plugin models are used to generate an image based on the condition image.

Parameters

context	an Android `ERROR(/Context)`.
generatorOptions	an `ImageGenerator.ImageGeneratorOptions` instance.
conditionOptions	an `ImageGenerator.ConditionOptions` instance.

Throws

	if there is an error during `ImageGenerator` creation.

public ImageGeneratorResult execute (boolean showResult)

Executes one iteration of image generation. The method must be called iterations times to generate the final image. Must call ERROR(/setInputs) before calling this method.

This is an iterative API, which must be called iteratively.

This API is useful for showing the intermediate image generation results and the image generation progress. Note that requesting the intermediate results will result in a larger latency. Consider using the e2e API instead for latency consideration.

Example usage:

imageGenerator.setInputs(prompt, iterations, seed); for (int step = 0; step < iterations; step++) { ImageGeneratorResult result = imageGenerator.execute(true); }