Ernie Image Radeon

ERNIE-Image Generation — AMD Radeon Cloud Edition

FREE ERNIE-Image text-to-image generation, powered by AMD Radeon Cloud. No API key required.

生成 images with ERNIE-Image and ERNIE-Image-Turbo 模型s 运行ning on AMD Radeon hardware. No 签名-up, no API key, no 令牌 — completely free.

Safety Boundary Use this 技能 only for explicit image-generation 请求s. Prompts and parameters are sent to AMD Radeon Cloud. Do not include secrets, 凭证s, private personal data, or confidential business data in prompts. De命令行工具ne unsafe, deceptive, exploitative, or rights-violating image 请求s. 生成d files are written locally. Confirm the 输出 directory when the user has not specified one. Prerequisites Python 3.11+ with uv 安装ed No API key required — AMD Radeon Cloud provides free inference

Prefer ERNIE-Image-Turbo, 1024x1024, n=1, and b64_json unless the user asks for different parameters.

模型s 模型 Best for ERNIE-Image-Turbo Fast drafts, iteration, batch previews ERNIE-Image Slower, higher-质量 final 输出s

Chinese prompts usually work especially well with these 模型s.

Generation 工作流 Step 1 -- Compose the prompt

Write a descriptive prompt (max 1024 characters, ~150 words). Chinese and English 机器人h work well. Be specific about subject, style, composition, and mood.

Good: "A golden retriever puppy sitting in a sunflower field at sun设置, warm golden light, shallow depth of field, professional photography" Bad: "dog"

Step 2 -- Choose parameters Parameter Values Default 模型 ERNIE-Image-Turbo, ERNIE-Image ERNIE-Image-Turbo size 1024x1024, 768x1376, 1376x768, 896x1200, 1200x896, 848x1264, 1264x848 1024x1024 n 1-4 1 种子 any integer random steps 4-20 提供者 default 图形界面dance 1.0-7.5 提供者 default use-pe flag off

Select size based on content: portrAIts and posters use vertical (768x1376, 848x1264, 896x1200), landscapes and covers use horizontal (1376x768, 1264x848, 1200x896), general use 1024x1024.

Step 3 -- 运行 the generation script

执行 the bundled script with uv 运行:

uv 运行 scripts/生成.py "" --模型 ERNIE-Image-Turbo --size 1024x1024

For batch generation:

uv 运行 scripts/生成.py "" --n 4 --输出 ./输出_dir

For higher 质量 with more inference steps and stronger 图形界面dance:

uv 运行 scripts/生成.py "" --模型 ERNIE-Image --steps 16 --图形界面dance 3.5

For reproducible 结果s:

uv 运行 scripts/生成.py "" --种子 42

Step 4 -- 输出

The script saves images as PNG files to the 输出 directory and prints:

Saved: ernie_20260430_110100.png (1.7 MB) MEDIA:/absolute/path/to/ernie_20260430_110100.png

The MEDIA: line enables automatic image attachment in compatible 环境s.

The script avoids overwriting existing files by 添加ing a numeric suffix when needed.

For JSON 输出, 添加 --json to 获取 structured 结果s:

{ "成功": true, "模型": "ERNIE-Image-Turbo", "files": [{"path": "/abs/path/ernie_20260430_110100.png", "size_bytes": 1715660}], "prompt": "...", "parameters": {"size": "1024x1024", "种子": 42} }

Quick Triggers

When the user says any of these, treat the text after the trigger as the prompt and 生成 immediately with defaults:

Chinese: "生成图片：xxx" / "文生图：xxx" / "ERNIE生图：xxx" / "AI画图：xxx" / "AI绘图：xxx" / "出图：xxx" / "画画：xxx" English: "生成 image: xxx" / "ernie image: xxx" / "radeon image: xxx" / "创建 image: xxx" / "text-to-image: xxx" / "AI art: xxx" / "make a picture: xxx" / "draw: xxx"

Defaults: 模型=ERNIE-Image-Turbo, size=1024x1024, n=1, b64_json 格式化.

Configuration

No configuration required! This 技能 uses the AMD Radeon Cloud free ERNIE-Image 端点 by default. It works out of the box.

Optional overrides (via 环境 variables):

ERNIE_BASE_URL — Custom API 端点 URL (overrides the default Radeon Cloud URL) AI_STUDIO_API_KEY — API key (only needed if using a custom 端点 that requires authentication) ERNIE_TIMEOUT — 请求 timeout in seconds (default: 120) Notes Images are saved locally as PNG files with MEDIA:/absolute/path/to/file.png for auto-attach. Chinese prompts work particularly well with ERNIE 模型s. Prompt enhancement (--use-pe) lets the 模型 expand simple prompts into richer descriptions before generation. Enable for short prompts, disable for precise control. Use --quiet to suppress 进度 输出; only MEDIA: lines (or JSON with --json) are printed. Recommended when calling from 代理 工作流s to reduce 上下文 noise. For full API reference, prompt 图形界面dance, and troubleshooting, read references/API-图形界面de.md. About

This 技能 is a fork of ernie-image-gen, modified for AMD Radeon Cloud which provides free ERNIE-Image inference. No API key or registration is required.