❄️ Azure Coldsnap
Azure Coldsnap is the anime and humanoid-focused counterpart to Azure Starfire.
This model emphasizes clean composition, stable generation, and balanced background detailing avoiding cluttered or overly complex scenes.
❄️ ColdSnap 3 Change Log
Major Changes
This update introduces a significant shift in model direction.
ColdSnap is no longer focused purely on anime humanoid.
It has been reworked into a universal base model, supporting both furry and humanoid
Base Model Merge (Major Update)
The base composition has been fully reworked with new core merges
🌿 Plant Milk — Model Suite (Walnut)
💫 WAI-illustrious-SDXL V15
🌟 Azure Starfire Void Stable V5
🎨 LoRA Integration
Added 3 LoRA merges
🧠 Base Model Merge V1-2
🌿 Plant Milk — Model Suite - Walnut | Illustrious Checkpoint
💫 AlcAI — AnimeHaven - Awakening - v10 | Illustrious Checkpoint
🌟 MAGNIFICENT — v1.0 | Illustrious Checkpoint
⚙️ Recommended Settings
Steps: 25–40
CFG Scale: 4–7 (values above 8 may cause oversharpened lighting)
Sampler: DPM++ 2M Karras / Euler a
Hires Fix: ✅ Recommended for character close-ups
FaceDetailer: Optional for portrait or semi-close shots
🏷️ Prompt Tags
✅ Positive (Quality Boost):
masterpiece, best quality, ultra-detailed, 8k, soft lighting, illustration, anime style, clean background, highly detailed eyes
🚫 Negative (Suggested):
lowres, (worst quality:1.2), (bad quality:1.2), (low quality:1.2), bad anatomy, deformed hands, overexposure
Description
FAQ
Comments (7)
bro, why is that so realistic? I added 3d, realistic in neg. but still
Did you put Photorealistic in your positive ?
@BastionExa nope, masterpiece and best quality only
@compgamer1337267 and the prompt + LoRa ?
In pos: anime coloring, {your prompt}
In neg: 3d, realistic, photorealistic, photo
Solid model I really like it! You just need to be a bit more detailed with your prompts but that’s not a bad thing model does a great job keeping two characters from blending their traits together.However the furryrelated details are a bit missing
ty that kinda the best part using Danbooru dataset it is very organized and clean on Tags but lacking furry side i have to merge e6 LoRa that have it