FLUX.2-pro Image Generation#

Generate high-quality images from text prompts using FLUX.2-pro on Azure AI Foundry.

Setup#

Set environment variables (typically in OpenClaw config or shell profile):

bash

export FLUX_ENDPOINT="https://<resource>.services.ai.azure.com/providers/blackforestlabs/v1/flux-2-pro?api-version=preview"
export FLUX_API_KEY="YOUR_KEY_HERE"

Quick Generation#

Use the bundled script:

bash

node scripts/generate.mjs --prompt "a red fox in autumn forest" --output fox.png
node scripts/generate.mjs --prompt "cute robot" --width 1440 --height 816 --output robot.png

Direct API Call#

When the script isn't suitable, call the API directly:

bash

curl -s -X POST "$FLUX_ENDPOINT" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $FLUX_API_KEY" \
  -d '{"prompt":"a cat","width":1024,"height":1024,"n":1,"model":"FLUX.2-pro"}' \
  | node -e "process.stdin.on('data',d=>{const j=JSON.parse(d);require('fs').writeFileSync('out.png',Buffer.from(j.data[0].b64_json,'base64'))})"

Critical parameters: "n": 1 and "model": "FLUX.2-pro" are mandatory. Omitting them causes HTTP 500.

Chinese Text Overlay#

FLUX cannot render CJK characters. Overlay text with Node.js Canvas:

javascript

import { createCanvas, loadImage, registerFont } from "canvas";
registerFont("NotoSansCJK-Bold.otf", { family: "NotoSansCJK" });

const img = await loadImage("base.png");
const canvas = createCanvas(img.width, img.height);
const ctx = canvas.getContext("2d");
ctx.drawImage(img, 0, 0);
ctx.font = 'bold 48px "NotoSansCJK"';
ctx.fillStyle = "#ffffff";
ctx.fillText("你好世界", 100, 100);

Requires: npm install canvas + a CJK font file (e.g., NotoSansCJK-Bold.otf).

Best Practices#

• Serialize requests — avoid parallel API calls; use sequential generation
• Set 180s timeout — generation can take 30–120 seconds
• Prompt in English — FLUX works best with English prompts
• Content filter — avoid violent, sexual, or otherwise filtered content in prompts
• Print quality — use 1240×1754 for A3, scale up as needed

API Reference#

See references/api.md for full request/response schema and size options.