1328 hand-tested AI prompts across image, video, and music — each with full context, variables, and a sample output. Content is authored in Traditional Chinese; the UI here is English.
A single line hides a macro-to-wide transition…
Pairing a timelapse with metamorphosis hands t…
Green-and-purple aurora, an icy reflection, an…
A single decisive action, dramatic side lighti…
A side-tracking shot with hooves kicking up du…
The instant of crossing the finish line, golde…
It locks straight onto three elements: the han…
Using in action and spinning explicitly demand…
Naming the highly recognizable nerikomi techni…
Melting evenly with a gentle flicker captures …
Short but clear on both the visual wax melt an…
Dripping wax forming delicate patterns plus na…
Six keywords lock the spaghetti-western look: …
Going straight for Sergio Leone's signature ex…
A massive dragon, a collapsing Gothic tower, d…
A crumbling tower, vines, an orange storm sky,…
An overpass swallowed by vegetation, with the …
Shattered glass towers after an EMP and glitch…
A lone survivor, a gas mask, a scorched desert…
Short and sharp, naming red, blue, and gold, i…
Paste in a whole report, article, or meeting minutes and it compresses the content straight into a single Traditional Chinese infographic. No design software needed — good for slide-deck visuals, long-form social posts, and internal communication; includes a built-in guard against inventing outside information.
Produces 4K e-commerce product shots without a physical studio: material words (matte finish, brushed texture) plus lighting description (softbox, fill light, shadow) are the key to a premium look. Good for listing images and homepage hero shots; fix small flaws with follow-up chat rather than regenerating.
No masks, no selection tools — just describe in plain language what to change. This prompt is built around the key declaration 'change only this one thing, keep everything else the same,' and the same sentence pattern covers changing backgrounds, removing passersby, or swapping outfits. If a result isn't quite right, just add another line to the same conversation instead of regenerating from scratch — much faster than starting over.
To generate a street scene that unmistakably reads as Taiwan, the key isn't writing 'Taiwan' — it's naming local elements like covered arcade walkways, scooters, overhead wires, and Traditional Chinese shop signs, then locking the sign text in quotes to prevent typos. Good for localized editorial imagery, presentation scenes, and social media backgrounds.