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.
PRO01{ "type": "evolutionary timeline infographic", "instruction": "Using R…
PRO02An anime-style illustration of a {argument name="action type" default="h…
PRO03{ "type": "live stream UI mockup", "subject": { "description": "port…
PRO04A JSON-style prompt template for an illustrated map infographic, with the visual style driven by an argument placeholder...
PRO05Visual design spec: 9:16 aspect ratio (vertical mobile infographic); background texture is a breathable beige handmade washi paper with subtle fiber grain…
PRO06A hand-drawn style city food map themed around Taizhou. The image uses a simplified bird's-eye-view hand-drawn city map as its base, labeling districts like Jiaojiang, Luqiao, and Huangyan, and waterway landmarks like the Ling River and Taizhou Bay, without aiming for precise scale…
A front-of-package label for a food or handmade brand, with the brand name in Chinese and English side by side, and the volume/weight and ingredient fine print rendered letter-perfect.
Midjourney v7 locks appearance with a reference image plus --sref for consistency, mass-producing a chibi mascot across multiple expressions and poses for LINE stickers or a brand IP character.
Turn your own voice into a reusable AI singer. Build a Voice profile from a 10-second sample, verify it with a Mandarin test phrase, and dial in the Audio Influence value to hit the sweet spot between "sounds like you" and "stays on pitch."
Taiwanese Hokkien vocals in Suno still frequently switch back to Mandarin mid-song. The fix is writing 'Hokkien pop' rather than '台語' in the Style field, using Custom Mode to lock in your own hand-written Chinese-character lyrics, then proofreading line by line. This is a pain point unique to Taiwan that the English-speaking Suno community doesn't cover at all.
A two-platform workflow: generate the main track in Suno, then fine-tune in Udio — use inpainting to select the 2 seconds where the enunciation is muddy or off-key, and regenerate only that section instead of re-rolling the whole song.
A three-act spread-to-explosion-to-reverse-reb…
One prompt cut into four shots, each with its …
Spelling out caustic patterns, dappled light r…
Splitting the JSON into fields locks in a fixe…
Adding the emotional hook of a person dropping…
Scaling unboxing up into a mechanical containe…
Using cool-light dull versus warm-light dewy d…
Emphasizing natural pores and a dewy glistenin…
A capsule enclosure, city landmarks, and tiny …
Anchor the wuxia style to Crouching Tiger, Hid…
Break the dragon transformation into a gradual…
Use timecoded three-part blocking (windup, lun…
Lock four parameters per shot: focal length (2…