Pirate Voyage
Redraw the person in an uploaded photo as the captain of an anime sea crew.
When this is the right skill
Pick this when someone wants themselves as an anime sea-adventure hero — a
ship deck, an open coat, tropical colour, big heroic energy.
Choose a different skill instead when the user asks for:
- a stealth/rooftop action look →
ninja-village
- themselves with a creature sidekick →
monster-tamer
- the house marker-and-ink doodle look →
doodle-avatar
- a moving version →
doodle-dash (an anime action animation)
- several styles at once →
style-roll
The naming rule (not optional)
Name the genre and the look, never a series, studio, artist or character —
in the prompt, in what you say to the user, anywhere. Two reasons, both real:
upstream image models refuse many named-character prompts, so the skill would
fail at random and look like our bug; and asking for a specific character’s
likeness produces a derivative of a protected character, which is the actual
legal exposure. “Anime sea-adventure captain” gets the recognition without it.
If the user names a show, that is fine — they can say what they like. Do not
repeat the name into the generation. Answer with the genre wording instead.
What to draw
- The same person: hairstyle, hair colour, face shape, skin tone, freckles,
glasses, facial hair all preserved.
- Anime facial grammar: large expressive eyes with a bright highlight, small
simplified nose, confident open smile. Full anime proportions, not chibi.
- Invented captain costume — long open coat, waist sash, loose shirt, cuffed
boots. No emblem, crest or flag mark on anything.
- Bright tropical palette, ship-deck props, an island on the horizon.
- Motion lines on the background with the figure kept sharp — the anime
convention, not streaks smeared across the character.
How to run it
- Confirm a photo of the person is attached. Without one there is no likeness
to preserve, so ask for one rather than inventing a face.
- Call
generateDoodle with skill: "pirate", the photo as imageUrl, and
any costume, colour or setting notes the user gave as description.
- Pass
refImageUrl when the message includes a separate style reference.
- Report the result briefly; the app renders the image itself.
Returns ONE image and costs 1 credit.
Following up
Offer one next step — the same person in the stealth-action look
(ninja-village), a creature companion (monster-tamer), or bringing this
scene to life as a moving doodle (doodle-dash).
