| Property | Value |
|---|---|
| Model Type | Text-to-Image Diffusion |
| License | Fair AI Public License 1.0-SD |
| Primary Task | Image Generation |
| Architecture | Stable Diffusion XL |
What is NONAMEmix_v1?
NONAMEmix_v1 is a sophisticated merged model combining several high-quality SDXL base models, specifically designed for generating aesthetic illustrations. It utilizes a unique merger recipe that includes Raehoshi illust XL 2.1, NoobAI-XL Epsilon-Pred, Movable Figure Model XL, and other components to create a versatile image generation system.
Implementation Details
The model implements a complex merging formula: A + (C - B) * 0.2 + (D - B) * 0.5 + E, where each component contributes specific artistic elements. It's optimized for various resolutions ranging from 640x1536 to 1536x640, with specific support for common aspect ratios.
- Optimized for CFG values between 5-6
- Recommended 28 sampling steps
- Uses DPM++2M SDE sampler
- Supports Hires.fix with 4x-UltraSharp upscaler
Core Capabilities
- High-quality illustration generation
- Booru-tag prompt style support
- Multiple resolution support
- Advanced upscaling capabilities
- Optimized for aesthetic image creation
Frequently Asked Questions
Q: What makes this model unique?
The model's unique strength lies in its carefully crafted merge recipe that combines multiple specialized models, optimized for aesthetic illustration generation with support for Booru-style prompting.
Q: What are the recommended use cases?
This model excels at generating high-quality illustrations, particularly when used with the recommended settings including masterpiece and best quality prompts, combined with appropriate negative prompts for optimal results.