@1365/stickerforge

StickerForge

AI-powered sticker pack generator. Turns any character image into a full set of emotion-variant stickers ready for messaging platforms.

How It Works

1. Trait Extraction — A vision model (Gemini 2.5 Flash) analyzes the input character image and extracts detailed visual traits: body color, flipper color, belly color, eye type, accessories, clothing, and an overall description.

2. Base Generation — An image generation model (Gemini 3 Pro) creates a full-body neutral-pose sticker using the extracted traits as strict guidance, ensuring character consistency.

3. Emotion Variants — The base sticker is edited into emotion variants (happy, sad, angry, surprised, love, cool). Each edit only changes expression and pose while preserving all character traits.

4. Post-Processing — Each sticker goes through:

   • Background removal (flood-fill from edges)
   • White die-cut sticker outline
   • Resize to platform specs
   • Format conversion (WebP for Telegram/WhatsApp, PNG for others)

5. Packaging — All stickers are packaged into a downloadable ZIP file.

Available Tools

sticker_from_pudgy

Generate a sticker pack directly from a Pudgy Penguin NFT — just provide the token ID. The tool automatically fetches the NFT image from IPFS (no image upload needed), retrieves on-chain metadata/traits, then runs the full sticker generation pipeline.

Parameters:

• token_id (required) — Pudgy Penguin token ID (0–8887)
• platform (optional, default: "telegram") — Target platform: telegram, whatsapp, discord, line, wechat
• emotions (optional, default: all 6) — List of emotions to generate: happy, sad, angry, surprised, love, cool

Returns: Path to the output directory containing individual stickers, sticker_pack.zip, the original NFT image, and both on-chain and vision-extracted traits.

Example usage:

• "I have Pudgy Penguin #6873, make me stickers" → sticker_from_pudgy(token_id=6873)
• "Make WhatsApp stickers for Pudgy #100, just happy and love" → sticker_from_pudgy(token_id=100, platform="whatsapp", emotions=["happy", "love"])

sticker_extract_traits

Extract visual traits from a character image. Returns a JSON object describing the character's colors, accessories, clothing, and overall appearance. Useful for understanding what the character looks like before generating stickers.

Parameters:

• image_path (required) — Path to the character image file (PNG, JPG, or WebP)

sticker_generate_pack

Generate a complete sticker pack from a character image. Runs the full pipeline: trait extraction, base generation, emotion variants, post-processing, and ZIP packaging.

Parameters:

• image_path (required) — Path to the character image file
• platform (optional, default: "telegram") — Target platform: telegram, whatsapp, discord, line, wechat
• emotions (optional, default: all 6) — List of emotions to generate: happy, sad, angry, surprised, love, cool

Returns: Path to the output directory containing individual stickers and a sticker_pack.zip.

sticker_generate_single

Generate a single emotion sticker from a character image. Useful for testing or generating one-off stickers.

Parameters:

• image_path (required) — Path to the character image file
• emotion (required) — One of: happy, sad, angry, surprised, love, cool
• platform (optional, default: "telegram") — Target platform

Returns: Path to the generated sticker file.

Supported Platforms

Platform	Size	Format	Max Size
Telegram	512px	WebP	512KB
WhatsApp	512px	WebP	100KB
Discord	320px	PNG	512KB
LINE	370px	PNG	1MB
WeChat	240px	PNG	1MB

Emotions

Emotion	Description
Happy	Big smile, jumping with raised flippers, sparkles
Sad	Droopy eyes, single tear, slouched pose
Angry	Furrowed brows, puffed cheeks, steam puffs
Surprised	Wide eyes, open mouth, flippers up, exclamation marks
Love	Heart eyes, blushing cheeks, floating hearts
Cool	Smug smirk, winking, peace sign, sparkles

Requirements

• OpenRouter API access is handled automatically via SC-Proxy — no API key configuration needed. All calls to openrouter.ai are routed through the proxy which injects credentials.
• Python packages: requests, Pillow (should already be installed)
• Uses core.http_client.proxied_post for all OpenRouter API calls