Verified against Nano Banana / Gemini 3.1 Flash Image · 2026-08-08
Create a costume portrait of a specific pet without losing what makes it recognizable
Composites an uploaded pet photo into a costume or scene concept while locking the specific markings, proportions, and expression that make that exact animal recognizable — so the result reads as "my actual dog" in costume, not a generic breed-template animal.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are creating a costume or themed portrait of a specific pet, using the uploaded photo as the source of truth for what this exact animal looks like. This is a costume added to a real, specific pet — not a generic illustration of their breed. PET REFERENCE a medium-sized corgi with a distinctive asymmetrical white patch over one eye and slightly larger left ear, photographed sitting, looking slightly off-camera. IDENTITY PRESERVATION Preserve exactly: this pet's specific coat pattern and markings (not a generic version of their breed's typical coloring), their actual proportions, ear shape and set, eye color, and any distinguishing features — a particular white patch, an asymmetrical marking, a specific scar or notch. Someone who knows this pet should recognize them immediately in the result, not see a generic stock-photo animal of the same breed wearing a costume. COSTUME OR SCENE CONCEPT a small knight costume — soft foam armor pieces and a tiny cape, sitting on a stone castle step. Fit the costume or scene elements onto this pet's actual body proportions and posture — a costume piece should drape or sit the way it actually would on an animal this specific size and shape, not on a generic idealized version of the breed. POSE AND EXPRESSION sitting proudly, head slightly tilted the same way as in the reference photo, alert and slightly curious expression. Keep the pose physically natural for how this specific animal actually holds itself — if the reference photo shows a dog with a particular head tilt or ear position that's part of their normal expression, that same character should carry through into the new pose rather than defaulting to a generic, breed-stock pose. BACKGROUND SETTING a softly lit stone castle courtyard, warm late-afternoon light. LIGHTING Light the pet and the costume or scene elements consistently — same light direction and color temperature across the whole frame, so the costume doesn't look like it was added under different lighting than the pet itself. REALISM AND CHARM BALANCE the white eye patch and slightly larger left ear are the two features people recognize this dog by — do not smooth these into symmetrical corgi markings. Keep the result looking like a real photograph of a real pet in costume, not an illustrated or overly stylized rendering, unless a specific illustrated style was actually requested — the whole appeal of this kind of portrait comes from it still reading as this actual animal, just dressed up. OUTPUT One portrait image where the pet is unmistakably the same specific animal from the reference photo, styled with the requested costume or scene, convincingly lit as one coherent photograph.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The instruction to preserve this pet's specific coat pattern and markings rather than a generic version of their breed's typical coloring targets a real and common drift in AI pet portraits: a model asked to render "a corgi in a knight costume" without a locking reference will often default toward the most statistically typical, idealized version of that breed's markings, which for an animal with a specific asymmetrical feature — an off-center eye patch, an oversized ear — produces a technically breed-accurate but personally wrong result, and the entire emotional value of a pet portrait for its owner rests on it being recognizably their specific, imperfect, individual animal rather than a breed-standard illustration. Second, carrying the pet's actual habitual expression and head-carriage characteristics into the new pose — rather than defaulting to a generic sit-and-look-at-camera pose — matters for the same recognition reason applied to behavior rather than appearance: a lot of what makes a specific pet feel like themselves in a photo is a characteristic head tilt or ear position that's part of how that individual animal actually carries itself, and losing that in favor of a generic, breed-stock pose is a second, separate way the result can technically be "the right dog" while still not feeling like their dog to the person who knows them best. Third, the realism-versus-stylization instruction matters because costume-portrait generation defaults easily toward an illustrated or overly polished look once a whimsical concept like a knight costume is introduced, and that stylization shift, however charming on its own, works against the same identity-preservation goal the rest of the brief is built around — a photorealistic treatment keeps the pet's actual markings and proportions rendered as real, specific detail, while a stylized illustration necessarily simplifies and generalizes those same features, which is exactly the trade-off this portrait is trying to avoid.
Verified against
Nano Banana / Gemini 3.1 Flash Image Gemini 3.1 Flash Image · 2026-08-08
Changelog
- 2026-08-08 — Initial publish, verified against Nano Banana (Gemini 3.1 Flash Image) on a corgi knight-costume portrait preserving asymmetrical markings.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
