This project provides a tool to generate custom Slack emojis from text prompts, leveraging AI for image creation. It's designed for users who want to quickly create unique emojis for communication platforms like Slack.

How It Works

The application utilizes a fine-tuned Stable Diffusion XL (SDXL) model for emoji generation, a prompt classifier (fine-tuned Llama-13b) to assess toxicity, and a background removal model. These components are orchestrated within a Next.js application, with data managed by PlanetScale (MySQL) and Vercel KV (Redis), and storage handled by Vercel Blob.

Quick Start & Requirements

• Install/Run: Deploy to Vercel using the provided button.
• Prerequisites: Requires setup for Replicate (AI provider), PlanetScale (database), Vercel Blob (storage), and Vercel KV (Redis).
• Links: Deploy to Vercel, GitHub

Highlighted Details

• Leverages fine-tuned SDXL for emoji generation based on Apple's emojis.
• Includes a prompt classifier for toxicity detection.
• Features background removal for generated images.
• Built with Next.js App Router and Server Actions.

Maintenance & Community

• Authors include Alexandru Ţurcanu and Dylan Player.
• Credits notable open-source projects and fine-tuned models.

Licensing & Compatibility

• License: AGPL-3.0 License.
• Compatibility: AGPL-3.0 is a strong copyleft license. Use in commercial or closed-source projects requires careful consideration of its terms, particularly regarding modifications and distribution.

Limitations & Caveats

The project is licensed under AGPL-3.0, which may impose significant obligations on derivative works and linked components, potentially restricting commercial use or integration into proprietary systems without open-sourcing those systems.