How to generate an image with @aws-sdk/client-bedrock-runtime

Question

After creating a BedrockRuntimeClient in typescript, I don't know how to invoke the model and send the command.

const client = new BedrockRuntimeClient({
        region: "us-east-1",
        apiVersion: '2023-09-30',
        credentials:{
            accessKeyId: process.env.AWS_ACCESS_KEY_ID ?? '',
            secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY ?? ''
        }
    });

Answer 1

You have to create first an input like this:

const input = { 
    contentType: 'application/json',
    accept: '*/*',
    modelId: 'stability.stable-diffusion-xl-v0',
    body: `{
        "text_prompts":[
            {
                "text":"A beautiful picture of a bird"
            }],
        "cfg_scale":10,
        "seed":0,
        "steps":50}`,
};

Then, create the command and send it:

const command = new InvokeModelCommand(input);
const response = await client.send(command);

(you have to do this in an async function)

Next, convert the response into a json to get the base64 image

const blobAdapter = response.body;    

const textDecoder = new TextDecoder('utf-8');
const jsonString = textDecoder.decode(blobAdapter.buffer);

try {
const parsedData = JSON.parse(jsonString);
    return parsedData.artifacts[0].base64
} catch (error) {
    console.error('Error parsing JSON:', error);
    return 'TextError';
}

Finally, you can do whatever you want with the base64, like put it in an <img> tag

<img src={`data:image/png;base64, ${base64_image}`}>

Here's the final code:

export async function functionName(prompt:string){   
    const client = new BedrockRuntimeClient({
        region: "us-east-1",
        apiVersion: '2023-09-30',
        credentials:{
            accessKeyId: process.env.AWS_ACCESS_KEY_ID ?? '',
            secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY ?? ''
        }
    });

    const input = { 
        contentType: 'application/json',
        accept: '*/*',
        modelId: 'stability.stable-diffusion-xl-v0',
        body: `{
            "text_prompts":[
                {
                    "text":"${prompt}"
                }],
            "cfg_scale":10,
            "seed":0,
            "steps":50}`,
    };

    const command = new InvokeModelCommand(input);
    const response = await client.send(command);

    const blobAdapter = response.body;    

    const textDecoder = new TextDecoder('utf-8');
    const jsonString = textDecoder.decode(blobAdapter.buffer);

    try {
    const parsedData = JSON.parse(jsonString);
        return parsedData.artifacts[0].base64
    } catch (error) {
        console.error('Error parsing JSON:', error);
        return 'TextError';
    }
}

I hope this helps someone.