Generate Images with Amazon Bedrock and Stable Diffusion XL using Java

Amazon Bedrock is a fully managed service that offers a choice of high-performing foundation models (FMs) from leading AI companies. Stable Diffusion XL, offered by Stability.ai, is a powerful text-to-image model capable of generating high-quality images from textual prompts. This article will guide you through the process of using Amazon Bedrock to invoke Stable Diffusion XL for image generation using the Java SDK.

Before you can start generating images, you need to set up the Bedrock Runtime client in your Java environment. This involves creating a `BedrockRuntimeClient` instance, configuring it with your AWS credentials and the desired AWS Region. Here's how you can do it: ```java import software.amazon.awssdk.auth.credentials.DefaultCredentialsProvider; import software.amazon.awssdk.regions.Region; import software.amazon.awssdk.services.bedrockruntime.BedrockRuntimeClient; // Create a Bedrock Runtime client in the AWS Region you want to use. // Replace the DefaultCredentialsProvider with your preferred credentials provider. var client = BedrockRuntimeClient.builder() .credentialsProvider(DefaultCredentialsProvider.create()) .region(Region.US_EAST_1) .build(); ``` Make sure you have the AWS SDK for Java 2.x added to your project dependencies.

The `InvokeModel` API requires a specific payload format that the Stable Diffusion XL model understands. This payload includes parameters like the text prompt, style preset, and seed. You can create a JSON string representing this payload using the following template: ```java var nativeRequestTemplate = """ { "text_prompts": [{ "text": "{{prompt}}" }], "style_preset": "{{style}}", "seed": {{seed}} }"""; // Define the prompt for the image generation. var prompt = "A stylized picture of a cute old steampunk robot"; // Get a random 32-bit seed for the image generation (max. 4,294,967,295). var seed = new BigInteger(31, new SecureRandom()); // Choose a style preset. var style = "cinematic"; // Embed the prompt, seed, and style in the model's native request payload. String nativeRequest = nativeRequestTemplate .replace("{{prompt}}", prompt) .replace("{{seed}}", seed.toString()) .replace("{{style}}", style); ``` Customize the `prompt`, `style`, and `seed` variables to generate different images.

With the Bedrock Runtime client set up and the request payload crafted, you can now invoke the Stable Diffusion XL model using the `invokeModel` method: ```java import software.amazon.awssdk.core.SdkBytes; import software.amazon.awssdk.core.exception.SdkClientException; // Set the model ID, e.g., Stable Diffusion XL v1. var modelId = "stability.stable-diffusion-xl-v1"; try { // Encode and send the request to the Bedrock Runtime. var response = client.invokeModel(request -> request .body(SdkBytes.fromUtf8String(nativeRequest)) .modelId(modelId) ); // Decode the response body. var responseBody = new JSONObject(response.body().asUtf8String()); // Retrieve the generated image data from the model's response. var base64ImageData = new JSONPointer("/artifacts/0/base64") .queryFrom(responseBody) .toString(); return base64ImageData; } catch (SdkClientException e) { System.err.printf("ERROR: Can't invoke '%s'. Reason: %s", modelId, e.getMessage()); throw new RuntimeException(e); } ``` This code sends the request to Amazon Bedrock and retrieves the response.

The response from the `invokeModel` method contains the generated image data in base64 format. You need to decode this data to display the image. The code snippet below shows how to extract the base64 image data from the response: ```java import org.json.JSONObject; import org.json.JSONPointer; // Decode the response body. var responseBody = new JSONObject(response.body().asUtf8String()); // Retrieve the generated image data from the model's response. var base64ImageData = new JSONPointer("/artifacts/0/base64") .queryFrom(responseBody) .toString(); ```

Once you have the base64 image data, you can display it using a suitable image display method. The original code uses a `displayImage` method. You'll need to implement this method based on your chosen UI framework or image library. A simple example might involve converting the base64 string to a byte array and then using an image library to display it. ```java // Assuming you have a displayImage method displayImage(base64ImageData); ```

Here's the complete Java code example for invoking Stable Diffusion XL on Amazon Bedrock: ```java import org.json.JSONObject; import org.json.JSONPointer; import software.amazon.awssdk.auth.credentials.DefaultCredentialsProvider; import software.amazon.awssdk.core.SdkBytes; import software.amazon.awssdk.core.exception.SdkClientException; import software.amazon.awssdk.regions.Region; import software.amazon.awssdk.services.bedrockruntime.BedrockRuntimeClient; import java.math.BigInteger; import java.security.SecureRandom; import static com.example.bedrockruntime.libs.ImageTools.displayImage; public class InvokeModel { public static String invokeModel() { // Create a Bedrock Runtime client in the AWS Region you want to use. // Replace the DefaultCredentialsProvider with your preferred credentials provider. var client = BedrockRuntimeClient.builder() .credentialsProvider(DefaultCredentialsProvider.create()) .region(Region.US_EAST_1) .build(); // Set the model ID, e.g., Stable Diffusion XL v1. var modelId = "stability.stable-diffusion-xl-v1"; // The InvokeModel API uses the model's native payload. // Learn more about the available inference parameters and response fields at: // https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-diffusion-1-0-text-image.html var nativeRequestTemplate = """ { "text_prompts": [{ "text": "{{prompt}}" }], "style_preset": "{{style}}", "seed": {{seed}} }"""; // Define the prompt for the image generation. var prompt = "A stylized picture of a cute old steampunk robot"; // Get a random 32-bit seed for the image generation (max. 4,294,967,295). var seed = new BigInteger(31, new SecureRandom()); // Choose a style preset. var style = "cinematic"; // Embed the prompt, seed, and style in the model's native request payload. String nativeRequest = nativeRequestTemplate .replace("{{prompt}}", prompt) .replace("{{seed}}", seed.toString()) .replace("{{style}}", style); try { // Encode and send the request to the Bedrock Runtime. var response = client.invokeModel(request -> request .body(SdkBytes.fromUtf8String(nativeRequest)) .modelId(modelId) ); // Decode the response body. var responseBody = new JSONObject(response.body().asUtf8String()); // Retrieve the generated image data from the model's response. var base64ImageData = new JSONPointer("/artifacts/0/base64") .queryFrom(responseBody) .toString(); return base64ImageData; } catch (SdkClientException e) { System.err.printf("ERROR: Can't invoke '%s'. Reason: %s", modelId, e.getMessage()); throw new RuntimeException(e); } } public static void main(String[] args) { System.out.println("Generating image. This may take a few seconds..."); String base64ImageData = invokeModel(); displayImage(base64ImageData); } } ```

For more information about Amazon Bedrock, the AWS SDK for Java, and Stable Diffusion XL, refer to the following resources: * [Amazon Bedrock Documentation](https://docs.aws.amazon.com/bedrock/latest/userguide/) * [AWS SDK for Java 2.x API Reference](https://sdk.amazonaws.com/java/api/latest/) * [AWS Code Examples Repository on GitHub](https://github.com/awsdocs/aws-doc-sdk-examples) * [Stable Diffusion XL Documentation](https://stability.ai/)