Prompt library
Maybe you don't want another tool. You have ChatGPT or Gemini open already, and you just need the right words to get your photo onto a shelf as a collectible. This page is that: working prompts, what they genuinely produce, where they break, and an honest read on when doing it yourself stops being the fast way.
Every prompt below follows the same skeleton that commercial generators use: subject with identity locked, product format, materials, lighting, and packaging art. Swap the details to taste. They are written for photo-based image generation (upload a photo, then paste the prompt) — text-only generators will invent a face, which is exactly the problem the last section is about.
Make yours — $3.99 · 3 collectible looks
Upload your photo, then paste: "Turn the person in this photo into a retail-style collectible action figure, still sealed in its cardboard-and-plastic packaging. Keep the exact same face, skin tone and expression as the photo — do not beautify or change the identity. Window box art with a fictional product name, dramatic product photography lighting, studio background, sharp focus, 1:1."
The identity sentence is the part people leave out, and its absence is why most DIY figures look like a stranger in your clothes. If the first render drifts, regenerate with the identity sentence moved to the very start — model attention is strongest early in the prompt.
"Turn the person in this photo into a holographic trading card character. Same face, same skin tone, no beautification. Foil border, fictional character class and stat lines, glossy card surface reflecting soft studio light, floating over a dark collector's display, 1:1."
Stat lines and class names are where general models go overboard — expect a few rounds of nudging before the text on the card stops being gibberish. That gibberish is normal: image models render letterforms as shapes, not words.
"Turn the person in this photo into a clear acrylic desk standee, standing on a wooden desk next to a coffee mug and a laptop. Same face and proportions as the photo, no beautification. Soft daylight from a window, shallow depth of field, product photo realism, 1:1."
The desk context is what sells the scale — without a familiar object next to it, the standee reads as a flat cutout. A mug, a keyboard, a plant: any of them does the job.
Three things wear people down. Face drift: general chat models were not built for identity preservation, so the figure comes back resembling a cousin. Daily limits: three good renders out of a dozen attempts is a normal hit rate, which burns through free tiers fast. And packaging art: getting window-box art that doesn't look like clipart takes patience most people don't have after the third try.
None of this is fatal — plenty of people enjoy the fiddling, and the prompts above give you a real head start. But if what you wanted was the figure, not the hobby, that's the moment the done-for-you route earns its $3.99: one upload, identity locked by the pipeline instead of by prompt luck, three styles delivered in about two minutes, refund if all three fail. Both routes are legitimate; know which one you're in the mood for.
The prompts themselves, yes — they're yours. But check your generator's terms on commercial use of outputs, and keep real brand names out of packaging art; a fictional product line (like the one our renders use) avoids trademark trouble entirely.
General image models trade identity for style. The fixes, in order of effectiveness: put the identity sentence first in the prompt, use a clear front-facing source photo, and regenerate. Our pipeline exists because locking identity is a solved problem when it's built in rather than requested.
They'll produce a figure — of an invented person. If you want a figure of yourself or someone specific, photo input is the only reliable path; text-only generation cannot know a face it has never seen.
No signup: the browser you pay from is your account. One payment, all three collectibles from one photo you already have.
Make yours — $3.99 · 3 collectible looks