AI Agent Hub
Back to skills
Seedream 5.0 Lite Image Generator icon

Seedream 5.0 Lite Image Generator

Design & Media Updated 2026.08.29

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_e942ebfc/seedream5-lite.

About this skill

Problem

Many teams want to call seedream 5.0 lite for image generation, but the workflow often breaks down on API authentication, reference image upload, task polling, and result download. This skill wraps the redfox.hk endpoint into a command-line tool, so a prompt can produce a downloadable image without manually handling the surrounding steps.

How It Works

It is organized around image generation tasks: provide a prompt, with optional --image reference, size, format, and group parameters. The script uploads local references, submits the task, polls the status, and downloads the result. Core capabilities include:
- Text to image: Chinese or English prompts generate PNG or JPEG output.
- Image to image: edit or restyle an existing reference image.
- Sequential sets: auto mode can generate up to 15 related images.
- High resolution: supports 2K/3K/4K or custom pixel sizes.
- Prompt optimization: standard and fast modes.
After completion, the output is saved as an image file that can be opened directly.

Boundaries and Caveats

It depends on redfox.hk and a valid API key; it is not a local inference tool. Typical generation takes 10 to 60 seconds, with longer times for complex requests. Polling runs for about 6 minutes at most, after which the task ID can be used for later lookup. Local reference images must be readable, and the output directory must be writable. It is useful for validating ideas, creating cover images, and generating batch sets, but unreviewed output should not be treated as final delivery.

Use Cases

  • Generate six 4K product cover candidates for Xiaohongshu posts using sequential image mode.
  • Send a local product cutout to image-to-image mode and restyle it as a café scene.
  • Draft a 2048x2048 UI concept from one English prompt to validate the visual direction.
  • Submit a batch task, move on, then retrieve and download the finished images by task ID.

Best For

  • Content operators creating social cover images and needing several consistent visual drafts from copy.
  • E-commerce operators turning local white-background product photos into unified scene images.
  • UI designers validating visual concepts with quick English prompt drafts.
  • Python engineers integrating image APIs and wanting command-line auth, polling, and downloads.