🧪 This application uses the JoyCaption model to generate high-quality, detailed captions for images.
Key Features:
- 🆓 Free & Open: No restrictions, open weights, training scripts included
- 🔓 Uncensored: Equal coverage of SFW and NSFW concepts
- 🌈 Diversity: Supports digital art, photoreal, anime, furry, and all styles
- 🎯 High Performance: Near GPT4o-level captioning quality
- 🔧 Minimal Filtering: Trained on diverse images for broad understanding
Supported image formats: PNG, JPG, JPEG, WEBP
Caption Styles:
- Formal Detailed: Long descriptive captions in formal tone
- Creative: Artistic and expressive descriptions
- Simple: Concise, straightforward descriptions
- Technical: Detailed technical analysis of composition and elements
- Custom: User-defined prompts for specialized captioning
Model: fancyfeast/llama-joycaption-alpha-two-hf-llava
Architecture: LLaVA with Llama 3.1 base