What it does

This is a personal prompt library by a Chinese AI practitioner who goes by Guizang. It collects their handcrafted prompts for image and video generation tools, organized into thematic recipes: holiday posters, 3D infographics, tilt-shift dioramas, even an “evil big banana” roast-comic generator. Most entries are .md or .json files you copy and adapt.

The interesting bit

The specificity is the point. These aren’t generic “make me a picture” prompts; they’re tuned for repeatable visual styles—Anthropic-style presentation decks, skeuomorphic glass cards, Nokia-era photo filters. It’s less “prompt engineering theory” and more “working chef’s notebook.”

Key highlights

• 16+ image prompt recipes, from utilitarian (screen-time visualization) to whimsical (miniature cities in glass bottles)
• One video prompt for first-frame-to-last-frame generation
• Bilingual Chinese/English README; prompts themselves are mostly Chinese
• Text section is currently empty (“Coming Soon”)
• 1,357 stars suggests a niche but engaged audience

Caveats

• No example outputs shown in the README—you’re trusting the prompts work as described
• No stated target model (Midjourney, DALL-E, Stable Diffusion, etc.); you’ll need to experiment
• “High quality” is self-claimed, not benchmarked

Verdict

Worth a browse if you’re stuck in a visual rut with generative AI and want style-specific starting points in Chinese. Skip it if you need tutorials, model-specific tuning guides, or English-language prompt logic.