WAI-illustrious-SDXL GGUF
quantized image generation model
1 min readupdated
This repository provides the popular WAI-illustrious-SDXL checkpoint (a model weights file for generation) converted to the GGUF format (a model weight compression format that saves memory). The model is optimized to run on GPUs with limited VRAM (video memory) without significant quality loss. It generates detailed images in anime and artistic styles. Built on the SDXL architecture, it is designed for use in interfaces like ComfyUI that support GGUF.
Key features
- Available in Q4_K_M, Q5_K_M, and Q8_0 quantizations for different memory budgets
- Optimized to run on GPUs with low VRAM
- Improved character accuracy and clean visual style
- Compatible with ComfyUI-GGUF nodes
What you can do
- Generate detailed anime-style illustrations and artwork in ComfyUI
- Run the model on PCs with modest hardware specs thanks to GGUF compression
- Generate at resolutions of 1024x1024 and higher to match native quality
Example
To run it in ComfyUI, place the model file in your ComfyUI/models/unet/ folder and use the recommended settings: CFG scale 4.0 - 6.5, Sampling steps 25 - 35, and the Euler a sampler.
Read more
DetailsFormats: Q4_K_M (~5-7GB), Q5_K_M (~8-10GB), Q8_0 (original size)
Requirements: requires ComfyUI-GGUF to be installed
Notes: VAE is baked into the model








![Ri-mix - [Illustrious + Anima]](/media/civitai--996495-ri-mix-illustrious-anima/cover-320.webp)
![Ri-mix - [Illustrious + Anima]](/media/civitai--996495-ri-mix-illustrious-anima/cover.webp)
Comments