Pet Halloween Costume · عوامی Open Skill
Pet Halloween Costume
Pet Halloween Costume Digen پر ایک عوامی Open Skill ہے۔ کھولیں اور دیکھیں یہ ورک فلو کیا تخلیق کرتا ہے۔
Auto
0 0 0
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4284862/2026-10-16/vjzm4n3j.jpg
# Pet Halloween Costume
Turns a reference photo into a festive Halloween costume portrait. The subject's identity — face, fur pattern, markings, eyes — stays exactly as in the uploaded photo; only the costume, pose, props, scene and lighting change to the Halloween theme.
This is a single-step skill: one `generate_image` call per requested variation. No multi-stage flow, no project state, no reference files.
## When to use
- The user uploads a pet photo and asks for a Halloween / costume / spooky-season portrait, or asks to "dress my pet up".
- The user wants an existing portrait from this conversation redone or adjusted (new costume, new pose, new background, different framing).
- The user has no photo and simply wants a Halloween pet portrait → go text-to-image.
## Steps
### 1. Pick the generation mode
| Situation | `model` |
|---|---|
| A photo of their pet is uploaded or already on the canvas | `i2i.hd.lite`, `i2i.hd` or `i2i.hd.gpt` (choose per the table below) |
| No photo provided, or the user asks for an invented/generic pet | `t2i` |
Image-to-image channel choice — always pass a channel, never a physical model name:
- `i2i.hd.lite` — fast, lower-fidelity pass. Use for quick previews, "show me a rough idea", or when the user is iterating through several costume ideas and will pick one to refine.
- `i2i.hd` — **default**. Balanced quality for a normal single-subject portrait.
- `i2i.hd.gpt` — highest prompt adherence and detail. Use when the user asks for the best/most accurate result, when the costume has specific details (patterns, props, accessories) that must be right, or when a previous attempt drifted from the photo.
### 2. Write the prompt (always English)
Fill this template — do not send a vague one-liner:
```
Festive Halloween portrait of {subject} with the exact same face, fur pattern, markings, eye color and identity as the reference photo; {costume}; {pose}; set in {scene}; {lighting}; {palette}; {composition}; {style}
```
Fill rules (pick ONE costume per generation):
- `{subject}` — "the pet in the reference photo" plus breed/name if the user gave one ("a fluffy grey tabby cat named Mochi"). Never describe a different animal or change the breed.
- `{costume}` — be specific, not "a Halloween costume": e.g. "a purple witch hat with a tiny black cape and a star-patterned collar", "a carved-pumpkin body suit with a green stalk hat", "a mini vampire cape with red satin lining and a bat-shaped bow tie", "a fuzzy ghost sheet with two eye holes", "a knitted bat-wing sweater with matching bat ears", "a spider-web lace collar with a small plastic spider".
- `{pose}` — e.g. "sitting upright facing the camera with its head slightly tilted", "standing on all fours with one front paw raised", "lying down on a cushion, chin resting on its paws".
- `{scene}` — e.g. "on a dark wooden floor surrounded by carved jack-o'-lanterns, dried autumn leaves and cobwebs", "in front of a foggy arched window with purple and orange bokeh string lights", "on a rustic porch step with hay bales and a glowing paper lantern".
- `{lighting}` — e.g. "warm pumpkin-glow key light from the left, cool purple rim light from behind, soft shadows".
- `{palette}` — e.g. "orange, purple and black Halloween palette with a touch of gold".
- `{composition}` — have the subject fill the frame: e.g. "centered portrait framing, subject fills the frame, shallow depth of field (85mm, f/2)".
- `{style}` — e.g. "photorealistic festive pet portrait, high detail on fur and fabric, no text, no watermark".
If a person appears in the photo (for example an owner holding the pet), keep them in the frame with the same face and identity and change only their costume/pose/setting — write that into `{subject}` explicitly.
Example filled prompt (cat, witch costume, floor scene):
```
Festive Halloween portrait of the pet in the reference photo, a fluffy grey tabby cat named Mochi, with the exact same face, fur pattern, whiskers, eye color and identity as the r