ERNIE-Image Atlas Cloud API + Batch Production: Enterprise AI Image Pipeline

Baidu Atlas Cloud provides ERNIE-Image API service — build enterprise-level batch image generation pipeline.

---

Atlas Cloud Platform Overview

Baidu AI Cloud Atlas provides ERNIE-Image cloud API service — call without local GPU.

Core Features

Feature	Description
Pay-per-use	Pay by generation count, no upfront cost
Elastic Scaling	Support high-concurrency batch generation
Global Acceleration	Global nodes, low latency
SLA Guarantee	99.9% availability

Access

• Console: https://console.bce.baidu.com
• API Docs: https://cloud.baidu.com/doc/ERNE.html

---

Basic API Call

Get API Key

1. Login to Baidu AI Cloud Console
2. Create application
3. Get Access Key and Secret Key

Python Example

import requests
import base64

def ernie_image_api(prompt, access_key, secret_key):
    url = "https://wenxin.baidu.com/v1/images/generations"
    
    headers = {
        "Authorization": f"Bearer {get_token(access_key, secret_key)}",
        "Content-Type": "application/json"
    }
    
    data = {
        "model": "ernie-image-8b",
        "prompt": prompt,
        "n": 1,
        "size": "1024x1024",
        "response_format": "b64_json"
    }
    
    response = requests.post(url, headers=headers, json=data)
    result = response.json()
    
    image_data = base64.b64decode(result["data"][0]["b64_json"])
    return image_data

def get_token(access_key, secret_key):
    url = f"https://aip.baidubce.com/oauth/2.0/token"
    params = {
        "grant_type": "client_credentials",
        "client_id": access_key,
        "client_secret": secret_key
    }
    return requests.get(url, params=params).json()["access_token"]

---

Batch Production Pipeline

Architecture

CSV/JSON input → Task queue → API call → Image storage → Output

Batch Generation Script

import csv
import requests
import base64
import time
from concurrent.futures import ThreadPoolExecutor

def batch_generate(csv_file, access_key, secret_key, output_dir="./output"):
    with open(csv_file, 'r', encoding='utf-8') as f:
        reader = csv.DictReader(f)
        tasks = list(reader)
    
    def generate_task(task):
        image_data = ernie_image_api(
            task['prompt'],
            access_key,
            secret_key
        )
        
        import os
        os.makedirs(output_dir, exist_ok=True)
        filename = f"{task['id']}.png"
        with open(f"{output_dir}/{filename}", 'wb') as f:
            f.write(image_data)
        
        return task['id']
    
    with ThreadPoolExecutor(max_workers=10) as executor:
        results = executor.map(generate_task, tasks)
    
    print(f"Generated {len(list(results))} images")

CSV Input Template

id,prompt,negative_prompt,width,height
1,"a cat on a table","","1024","1024"
2,"a dog in a garden","","1024","1024"
3,"a bird in a tree","","1024","1024"

---

ComfyUI + API Hybrid Pipeline

Architecture

Local ComfyUI (complex tasks) ↔ Atlas API (simple tasks)

Task Routing Logic

Task Type	Location	Reason
Simple generation	Atlas API	Low cost, fast
IP-Adapter	Local ComfyUI	Not supported by API
ControlNet	Local ComfyUI	Not supported by API
Inpainting	Local ComfyUI	Fine control
Batch production	Atlas API	Elastic scaling

---

Enterprise Deployment Options

Option 1: Full Cloud

Advantages:
- Zero hardware cost
- Elastic scaling
- Low maintenance cost

Cost:
- API calls: $0.001/image
- Storage: $0.01/GB/month

Option 2: Hybrid Cloud

Advantages:
- Core data local
- Elastic cloud scaling
- Flexible cost control

Cost:
- Local GPU: $7-14K
- API calls: Pay-per-use

---

Summary

Atlas Cloud API + Batch Production key points:

1. API call: Simple integration, pay-per-use
2. Batch pipeline: CSV/JSON-driven + concurrent execution
3. Hybrid deployment: Complex tasks local, simple tasks cloud
4. Enterprise-grade: SLA guarantee + global acceleration

Master Atlas Cloud API, and ERNIE-Image becomes your enterprise AI image engine.

---

Based on Baidu AI Cloud Atlas + ERNIE-Image API.