SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsimage-gen

image-gen

Design & creation

View repository

Updated 10-15 20:28

About this skill

AI image generation

A text-to-image tool based on the ModelScope API, supporting asynchronous generation and LoRA style stacking.

Configuration

Edit skills/image-gen/scripts/config.json：

{
  "api_token": "YOUR_MODELSCOPE_TOKEN"
}

Token acquisition:ModelScope Console Create an Access Token.

Command-line invocation

# 基础文生图
python3 skills/image-gen/scripts/generate.py generate --prompt "一只金色的猫"

# 指定输出路径
python3 skills/image-gen/scripts/generate.py generate --prompt "赛博朋克城市" --output cyberpunk.jpg

# 指定模型
python3 skills/image-gen/scripts/generate.py generate --prompt "水墨山水画" --model Tongyi-MAI/Z-Image-Turbo

# 使用 LoRA 风格
python3 skills/image-gen/scripts/generate.py generate --prompt "一个女孩" --lora "repo-id/lora-model"

# 多 LoRA 叠加（JSON 格式，权重之和为 1.0）
python3 skills/image-gen/scripts/generate.py generate --prompt "风景画" --lora '{"lora1": 0.6, "lora2": 0.4}'

# 指定尺寸
python3 skills/image-gen/scripts/generate.py generate --prompt "壁纸" --size 1920x1080

# 查询任务状态
python3 skills/image-gen/scripts/generate.py status --task-id TASK_ID

# JSON 格式输出
python3 skills/image-gen/scripts/generate.py generate --prompt "猫" --json

AI invocation flow

Step 1: Generate the image

Construct an English prompt based on the user's description (English works better), call the script to generate the image and save it locally:

python3 skills/image-gen/scripts/generate.py generate --prompt "A cute golden cat, high quality, detailed fur, studio lighting" --output data/temp/images/golden_cat.jpg

Step 2: Send to the channel (automatically links with send_media)

After the image is generated successfully, if the current session is a channel session (Feishu/QQ/DingTalk/Telegram, etc.), you must use the built-in tool send_media to send the image to the user:

send_media(file_paths=["data/temp/images/golden_cat.jpg"], message="已为你生成图片 🎨")

If it is a web session, just tell the user the path where the image is saved.

Complete example

The user says "draw me a cat":

  1. Execute generation:
python3 skills/image-gen/scripts/generate.py generate --prompt "A cute golden cat, high quality, detailed fur" --output data/temp/images/cat.jpg
  1. Send to the channel:
send_media(file_paths=["data/temp/images/cat.jpg"], message="🎨 已生成图片：一只金色的猫")

The user says "draw a cyberpunk-style city wallpaper":

  1. Execute generation:
python3 skills/image-gen/scripts/generate.py generate --prompt "Cyberpunk city skyline, neon lights, rain, cinematic, 4K" --output data/temp/images/cyberpunk.jpg --size 1920x1080
  1. Send to the channel:
send_media(file_paths=["data/temp/images/cyberpunk.jpg"], message="🎨 赛博朋克城市壁纸")

Supported models

Model Description
Tongyi-MAI/Z-Image-Turbo Default, Tongyi Wanxiang fast generation

You can specify other text-to-image models on ModelScope via the --model parameter.

Prompt tips

  • Using English prompts works better
  • Add quality descriptors:high quality, detailed, 4K, studio lighting
  • Add style descriptions:oil painting, watercolor, anime style, photorealistic
  • Add composition descriptions:close-up, wide angle, bird's eye view
  • Negative descriptions can be expressed in the prompt with no xxx express

Notes

  • Asynchronous generation; the script will automatically poll until completion (default timeout 5 minutes)
  • It is recommended to save generated images to the data/temp/images/ directory
  • LoRA stacking is limited to 6, and the sum of weights must be 1.0
  • After generating an image in a channel session, be sure to call send_media to send it; do not just return the file path