ASTX-MOSS

🎨 About this Model

A fully merged-tuned model. All aspects of the model have been improved: better prompt comprehension, image quality, attention to detail. The model has been tested with standard parameters.

The model creates images in various styles, from realism and 3D to pixel art; anime is optimized. The model is the best replacement option for Anima.

• 🧬 Base: created on a MOSS-Base model merge, not SFT

• 🔞 Can produce NSFW content

• ⚡ A quantized version will be uploaded, but a bit later

⚙️ Settings

🎬 My settings for MOSS

• 📸 CLIP Skip: 1 or None

• 🎲 CFG Scale: 3 – 3.5

• 🧼 Sampler / Scheduler: Res_Multistep or Euler with SGM_Uniform or Simple ⭐ best way!

• 👣 Steps: 25 – 35

• ♨️ Hi-Res Fix: ❌ I am not using it

• 🎭 ADetailer: ❌ I am not using it

📐 Recommended Resolution

• 1024×1536 / 1536×1024

• 832×1216 / 1216×832

• Simple: 1024×1024

✨ Prompt Parameters

✔️ Quality (MOSS): Natural Language and Booru Tags — no need for quality tags here.

❌ Negative (MOSS):

NONE