Adobe Core Workflow A — Firefly Services

Overview

Primary creative workflow using Adobe Firefly v3 APIs: text-to-image generation, generative fill (inpainting), and image expansion (outpainting). These are the most common Firefly Services operations for marketing asset automation.

Prerequisites

• Completed adobe-install-auth with Firefly API scopes (firefly_api,ff_apis)
• @adobe/firefly-apis installed, or direct REST access
• Pre-signed cloud storage URLs for input/output images (S3, Azure Blob, or Dropbox)

Instructions

Step 1: Text-to-Image Generation (Synchronous)

// src/workflows/firefly-generate.ts
import { getAccessToken } from '../adobe/client';

interface FireflyGenerateOptions {
  prompt: string;
  negativePrompt?: string;
  width?: number;    // 1024, 1472, 1792, 2048
  height?: number;
  n?: number;        // 1-4 images
  contentClass?: 'art' | 'photo';
  style?: {
    presets?: string[];  // e.g., ['digital_art', 'cinematic']
    strength?: number;   // 0-100
  };
}

interface FireflyOutput {
  outputs: Array<{
    image: { url: string };
    seed: number;
  }>;
}

export async function generateImage(opts: FireflyGenerateOptions): Promise<FireflyOutput> {
  const token = await getAccessToken();

  const body: Record<string, any> = {
    prompt: opts.prompt,
    n: opts.n || 1,
    size: { width: opts.width || 1024, height: opts.height || 1024 },
    contentClass: opts.contentClass || 'photo',
  };

  if (opts.negativePrompt) body.negativePrompt = opts.negativePrompt;
  if (opts.style?.presets) {
    body.styles = { presets: opts.style.presets };
  }

  const response = await fetch('https://firefly-api.adobe.io/v3/images/generate', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${token}`,
      'x-api-key': process.env.ADOBE_CLIENT_ID!,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify(body),
  });

  if (!response.ok) {
    const err = await response.text();
    throw new Error(`Firefly generate failed (${response.status}): ${err}`);
  }

  return response.json();
}

Step 2: Async Generation (for High Volume)

// For production pipelines, use async endpoint to avoid HTTP timeouts
export async function generateImageAsync(opts: FireflyGenerateOptions) {
  const token = await getAccessToken();

  const response = await fetch('https://firefly-api.adobe.io/v3/images/generate-async', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${token}`,
      'x-api-key': process.env.ADOBE_CLIENT_ID!,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      prompt: opts.prompt,
      n: opts.n || 1,
      size: { width: opts.width || 1024, height: opts.height || 1024 },
    }),
  });

  const { jobId, statusUrl, cancelUrl } = await response.json();
  console.log(`Firefly async job: ${jobId}`);

  // Poll for completion
  let result: any;
  while (true) {
    await new Promise(r => setTimeout(r, 2000));
    const poll = await fetch(statusUrl, {
      headers: {
        'Authorization': `Bearer ${token}`,
        'x-api-key': process.env.ADOBE_CLIENT_ID!,
      },
    });
    result = await poll.json();
    if (result.status === 'succeeded' || result.status === 'failed') break;
  }

  if (result.status === 'failed') throw new Error(`Async generation failed: ${result.error}`);
  return result;
}

Step 3: Generative Fill (Inpainting)

// Fill a masked region of an image with AI-generated content
export async function generativeFill(
  imageUrl: string,
  maskUrl: string,
  prompt: string
): Promise<FireflyOutput> {
  const token = await getAccessToken();

  const response = await fetch('https://firefly-api.adobe.io/v3/images/fill', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${token}`,
      'x-api-key': process.env.ADOBE_CLIENT_ID!,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      image: { source: { url: imageUrl } },
      mask: { source: { url: maskUrl } },
      prompt,
      n: 1,
    }),
  });

  if (!response.ok) throw new Error(`Fill failed: ${response.status}`);
  return response.json();
}

Step 4: Image Expansion (Outpainting)

// Expand an image to a larger canvas size with AI-generated surroundings
export async function expandImage(
  imageUrl: string,
  targetWidth: number,
  targetHeight: number,
  prompt?: string
): Promise<FireflyOutput> {
  const token = await getAccessToken();

  const response = await fetch('https://firefly-api.adobe.io/v3/images/expand', {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${token}`,
      'x-api-key': process.env.ADOBE_CLIENT_ID!,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      image: { source: { url: imageUrl } },
      size: { width: targetWidth, height: targetHeight },
      ...(prompt && { prompt }),
      n: 1,
    }),
  });

  if (!response.ok) throw new Error(`Expand failed: ${response.status}`);
  return response.json();
}

Output

• AI-generated images from text prompts (sync or async)
• Inpainted regions via generative fill with mask
• Expanded/outpainted images to larger canvas sizes
• Temporary URLs for generated images (download within 24h)

Error Handling

Examples

Start with the smallest applicable command or code example already provided in this guide, using a non-production Adobe environment and credentials. Confirm the documented response or validation result before applying the pattern to production.

Resources

• Firefly API Reference
• Firefly Generate Image Tutorial
• Using Async APIs

Next Steps

For PDF document workflows, see adobe-core-workflow-b.