🎨 Crazyrouter AI 图像生成指南

通过 Crazyrouter API 调用 DALL-E 3、Flux 等图像生成模型

Crazyrouter 不仅支持文本模型，还支持图像生成模型。一个 API Key 即可调用多种 AI 绘画模型。

支持的图像模型

模型	提供商	特点
`dall-e-3`	OpenAI	高质量、理解复杂提示词
`dall-e-2`	OpenAI	快速、便宜
`flux-schnell`	Black Forest Labs	开源、速度快
`flux-dev`	Black Forest Labs	高质量开源模型

快速开始

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://crazyrouter.com/v1",
    api_key="sk-your-crazyrouter-key"
)

response = client.images.generate(
    model="dall-e-3",
    prompt="A cute robot serving coffee in a futuristic cafe, digital art style",
    size="1024x1024",
    quality="standard",
    n=1,
)

image_url = response.data[0].url
print(f"Image URL: {image_url}")

cURL

curl https://crazyrouter.com/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-your-crazyrouter-key" \
  -d '{
    "model": "dall-e-3",
    "prompt": "A serene Japanese garden with cherry blossoms, watercolor painting",
    "size": "1024x1024",
    "n": 1
  }'

Node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://crazyrouter.com/v1",
  apiKey: "sk-your-crazyrouter-key",
});

const response = await client.images.generate({
  model: "dall-e-3",
  prompt: "A cyberpunk cityscape at night with neon lights, 4K detailed",
  size: "1024x1024",
  n: 1,
});

console.log(response.data[0].url);

DALL-E 3 详细用法

尺寸选项

尺寸	用途
`1024x1024`	正方形（默认）
`1024x1792`	竖版（手机壁纸、海报）
`1792x1024`	横版（桌面壁纸、Banner）

质量选项

质量	说明
`standard`	标准质量（默认，更快）
`hd`	高清质量（更细腻，更慢）

风格选项

风格	说明
`vivid`	鲜艳风格（默认）
`natural`	自然风格

完整示例

response = client.images.generate(
    model="dall-e-3",
    prompt="A majestic dragon flying over a medieval castle at sunset",
    size="1792x1024",    # 横版
    quality="hd",         # 高清
    style="vivid",        # 鲜艳
    n=1,
)

批量生成

prompts = [
    "A minimalist logo for a tech startup, blue and white",
    "A cozy reading nook with warm lighting, illustration style",
    "An astronaut playing guitar on the moon, digital art",
    "A steampunk mechanical butterfly, detailed 3D render",
]

for i, prompt in enumerate(prompts):
    response = client.images.generate(
        model="dall-e-3",
        prompt=prompt,
        size="1024x1024",
        n=1,
    )
    print(f"Image {i+1}: {response.data[0].url}")

下载图片

import requests

response = client.images.generate(
    model="dall-e-3",
    prompt="A beautiful sunset over the ocean",
    size="1024x1024",
    n=1,
)

image_url = response.data[0].url

# 下载保存
img_data = requests.get(image_url).content
with open("sunset.png", "wb") as f:
    f.write(img_data)
print("Image saved as sunset.png")

在客户端中使用

Cherry Studio

Cherry Studio 内置 AI 绘画功能，配置 Crazyrouter 后可以直接在客户端中生成图片。

ChatBox

ChatBox 支持图片生成，在对话中输入绘画提示词即可。

NextChat / LobeChat

这些客户端支持 DALL-E 插件，配置 Crazyrouter API 后可以在对话中生成图片。

提示词技巧

基础结构

[主体] + [风格] + [细节] + [氛围]

示例

提示词	效果
`A cat wearing a space suit, digital art, highly detailed`	太空猫，数字艺术风格
`Minimalist mountain landscape, Japanese ink wash painting`	极简山水，水墨画风格
`Product photo of a sleek smartphone, studio lighting, white background`	产品摄影风格
`Isometric view of a tiny coffee shop, pixel art, warm colors`	等距像素风咖啡店

风格关键词

写实: photorealistic, 8K, detailed, professional photography
插画: illustration, digital art, concept art, artstation
水彩: watercolor painting, soft colors, artistic
像素: pixel art, retro, 16-bit, game style
3D: 3D render, octane render, blender, cinema 4D
动漫: anime style, manga, studio ghibli
极简: minimalist, clean, simple, flat design

与文本模型结合

让 AI 先生成提示词，再用来绘画：

# Step 1: 用文本模型生成优化的提示词
chat_response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{
        "role": "user",
        "content": "Generate a detailed DALL-E prompt for: a cozy winter cabin. Include style, lighting, and mood details. Output only the prompt."
    }]
)
optimized_prompt = chat_response.choices[0].message.content

# Step 2: 用优化后的提示词生成图片
image_response = client.images.generate(
    model="dall-e-3",
    prompt=optimized_prompt,
    size="1792x1024",
    quality="hd",
    n=1,
)
print(f"Prompt: {optimized_prompt}")
print(f"Image: {image_response.data[0].url}")

ComfyUI 集成

如果你使用 ComfyUI 进行更高级的图像生成工作流，可以通过 Crazyrouter API 获取文本提示词优化：

# 用 Crazyrouter 的文本模型优化 ComfyUI 的提示词
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{
        "role": "system",
        "content": "You are a Stable Diffusion prompt expert. Convert user descriptions into optimized SD prompts with positive and negative prompts."
    }, {
        "role": "user",
        "content": "A beautiful girl in a flower garden"
    }]
)
print(response.choices[0].message.content)
# Output: positive prompt + negative prompt for ComfyUI/SD

价格参考

模型	尺寸	质量	价格/张
DALL-E 3	1024x1024	standard	~$0.04
DALL-E 3	1024x1024	hd	~$0.08
DALL-E 3	1792x1024	standard	~$0.08
DALL-E 3	1792x1024	hd	~$0.12
DALL-E 2	1024x1024	-	~$0.02

通过 Crazyrouter 调用，价格可能更优惠。

链接

🌐 Crazyrouter — 获取 API Key
🤖 在线 Demo
📖 快速入门
💻 编程工具配置
💬 Telegram 社区
🐦 Twitter @metaviiii

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support