Fashion Designer Creator · Open Skill 公開技能

Fashion Designer Creator

Fashion Designer Creator 是 Digen 上的公開 Open Skill。開啟後即可了解此工作流能製作什麼。

@jeeai

Series

6 2 5

https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/1821775/2026-09-21/6ggjv69d.jpg

# Fashion Designer Creator

## When to use
- The user wants a realistic, creative fashion-design video or package where a **reference person** acts as the fashion designer — designing, sketching, constructing, styling, or presenting an original garment, outfit, costume, accessory, or collection.
- Covers: fashion sketches, fabric selection, pattern-making, measuring and cutting, sewing and stitching, draping, embroidery and embellishment, sustainable fashion, traditional/cultural clothing, streetwear, modest fashion, formalwear, fantasy costumes, funny fashion designs, runway presentation.
- Also when they want to revise/continue an existing fashion-designer package, or generate just the thumbnail, voiceover, or video clips.

## Reference Person Rules (always preserve)
- Preserve the person's face, identity, hairstyle, skin tone, age, and body proportions — never change them into a different person.
- Show the person designing or making the clothing; keep hands, scissors, measuring tape, fabric, sewing machine, and tools realistic.
- Use practical designer clothing and a suitable studio environment.
- Do not copy the appearance of a real designer or celebrity.

## Steps
1. First use `write_todos` to list the steps and update it as you progress.
2. Clarify the request: which fashion category, whether they have a reference person image, and whether they want the full package, just the plan, or only a specific asset (thumbnail / voiceover / clips). If a reference image is provided, use it as the person.
3. Produce the full creative package (see "Required Output" below) and save it as a document with `create_document(content=..., name="Fashion Designer Package", entity_type="scripts")`.
4. Generate the assets (see "Generation" below). Use `set_guidance` after each stage to suggest next actions.
5. On success, summarize in natural language; never paste large URL blocks.

## Required Output (fill all fields)
```
PROJECT TITLE:
FASHION CATEGORY:
GARMENT OR OUTFIT:
TARGET AUDIENCE:
MOOD:
STYLE:
COLOUR PALETTE:
FABRIC:
DESIGN CONCEPT:

REFERENCE PERSON:
- Role:
- Clothing:
- Pose:
- Expression:
- Activity:

DESIGN PROCESS:
1. Inspiration and sketching.
2. Fabric and colour selection.
3. Measurements and pattern creation.
4. Cutting and preparation.
5. Sewing or assembly.
6. Decoration and finishing.
7. Final outfit presentation.

SCENE PLAN:
Scene 1:
- Location:
- Action:
- Camera:
- Materials:
- Narration:
- On-screen text:
- Generation prompt:

FINAL IMAGE OR VIDEO PROMPT:
NEGATIVE PROMPT:
THUMBNAIL TEXT:
THUMBNAIL PROMPT:
YOUTUBE TITLE:
DESCRIPTION:
KEYWORDS:
HASHTAGS:
```

## Generation
Use these channels (never hard-code physical model names). Prompts in English.

### Images (`generate_image`)
- **Fashion sketches / moodboard / fabric swatches / pattern diagrams** (no reference person): `model='t2i'`, `orientation='portrait'`, `name="Fashion Sketch"`, `entity_type="design_sketches"`.
- **Reference person in the studio / styling the outfit / runway presentation** (has a reference image): `model='i2i'`, pass the person's image in `input_urls`, `orientation='portrait'`, `name="Designer Scene"`, `entity_type="designer_scenes"`.
- **Thumbnail**: `model='t2i'` (or `i2i` if using the person), `orientation='landscape'`, `name="Thumbnail"`, `entity_type="thumbnails"`.

### Video (`generate_video`)
- **Reference person designing/making the garment** (has a reference image): `model='ref2v'`, pass the person's image in `input_urls`, `orientation='portrait'`, `name="Designer Clip"`, `entity_type="storyboard_videos"`. Describe the action (sketching, cutting, sewing, draping, presenting) in the prompt.
- **Pure text-to-video scene** (no reference image): `model='t2v'`, `orientation='portrait'`, `name="Designer Clip"`, `entity_type="storyboard_videos"`.
- Video generation does not support NSFW/adult content on any channel.

### Voiceover (`generate_tts`)
- Narration for the design process: `model='tts.design'`, `text=<narration>`, `voice_instruct