Dip-Pen Sketch Color
Reblis/dip-pen-sketch-color-skill
Converta fotos em esboços estilo Philly com marcador solto sobre papel branco
Ver no GitHubSkill do Claude Code que transforma fotos em esboços estilo dip-pen (nanquim) com cores soltas de marcador. Usa o modelo Nano Banana 2 do Google via fal.ai para edição de imagens com preservação de semelhança e remoção de fundo automática.
Instalação
npx convez-ai Reblis/dip-pen-sketch-color-skill1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Criar artes estilo TikTok/CapCut (Philly sketch look)
- Transformar retratos e fotos em ilustrações manuscritas
- Gerar variações artísticas de imagens com estilo retro
Veredito

Skill bem construída para um efeito visual específico (Philly sketch). Requer créditos fal.ai (~$0.03–0.04 por imagem) e dependências externas (ImageMagick, curl), o que pode desafiar iniciantes, mas o resultado é artisticamente autêntico e as instruções são claras.
README
A tradução para o português deste README ainda está na fila.
dip-pen-sketch-color — a Claude Code skill by Reblis.com
Turn any photo into a rough old-school dip-pen ink sketch with loose marker color:
dip-pen-sketch-color /path/to/photo.jpg

(The right panel is what this skill produces. The middle panel is the ink-only sibling.)
The "Philly sketch" TikTok/CapCut look: scratchy dip-pen, nib-and-India-ink linework with loose alcohol-marker (Copic-style) color streaked on top — flat, patchy fills that bleed outside the lines — and the subject lifted onto blank sketchbook paper. Deliberately rough, not a clean illustration.
Install
git clone https://github.com/Reblis/dip-pen-sketch-color-skill ~/.claude/skills/dip-pen-sketch-color
Restart Claude Code (or start a new session) so the skill registers, then run it on any photo path or image URL.
Requirements
This skill calls an AI image-editing model — it is not local/free:
- Claude Code
- A connected fal.ai MCP server with credits. The skill runs
fal-ai/nano-banana-2/edit— Google's Nano Banana 2 image-editing model, served through fal.ai. It's the part that preserves the subject's likeness while restyling. Cost ≈ $0.03–0.04 per image. curland ImageMagick (convert) for fetching and resizing.- Optional, free: a connected nanobanana / Gemini MCP runs the same Nano Banana model family at no cost — but its free tier is frequently rate-limited, which is why fal.ai is the default path.
What you get
- One image — ink sketch with loose marker color on blank white paper.
- Background removed — just the subject on empty paper.
- Likeness preserved — faces, expressions, hair, and clothing stay recognizable.
- Authentic marker feel — limited palette, streaky fills bleeding past the ink.
Related skills
- dip-pen-sketch-outline — the same look in pure black ink, no color.
- dip-pen-sketch-combo — outputs both outline and color with stroke-for-stroke matching lines.
Inputs
A local image path, a direct image URL, or — for Instagram/Facebook/Pinterest links (web pages, not images) — the skill downloads the photo first.
Credit
Built by Reblis.com. Uses Google's Nano Banana 2 via fal.ai. MIT licensed.