README.md

metadata

language:
  - en
pipeline_tag: text-to-image
library_name: diffusers
tags:
  - text-to-image
  - stable-diffusion
  - safetensors
  - anime
  - merge
license: openrail++
base_model:
  - neta-art/Neta-Lumina
  - hanzogak/comradeshipLUv2

LumiNetaMix

LumiNetaMix - Merge builds on the open-source Neta‑Lumina checkpoint using a custom karcher_mean merge. It preserves the original anime‑style aesthetic while delivering enhanced NSFW support and slight gains in generation quality—especially with simple and Danbooru tags.

• View on CivitAI
• Download LumiNetaMix (All‑in‑One)

• Original Model
  Building on the open-source Neta‑Lumina checkpoint.
• Prompt Book
  Full prompt guidelines and tag list: neta_lumina_prompt_book
• ComfyUI Workflow
  Pre-made workflow for ComfyUI:
  Download JSON

Model Details

• Model type: diffusion-based text-to-image
• Base checkpoint: neta-art/Neta-Lumina
• Merge method: karcher_mean custom merge

Key Improvements

• Enhanced NSFW support

• Slight quality gains with both simple and Danbooru prompt tags

• Maintains original anime aesthetic and multilingual tag comprehension

• Recommended Settings

  • Sampler: res_multistep | eular_ancestral
  • Scheduler: linear_quadratic
  • Steps: 30
  • CFG Scale: 4.0 – 5.5
  • Resolution: ≥ 1024×1024

Downloads

• All‑in‑One (AIO) Version
  Download the AIO model (UNET + Text Encoder + VAE)

• Separate Components

  • UNET checkpoint: Download UNET
  • Text Encoder: Download Text Encoder

Merge Sources

• Neta Lumina
  https://huggingface.co/neta-art/Neta-Lumina

• Comradeship LU v2
  https://huggingface.co/hanzogak/comradeshipLUv2

• Netayume Lumina (v2.0)
  CivitAI – Netayume Lumina (modelVersionId: 2026566)