🌟 About This Model: Precision & Clarity
Let's be honest: a lot of models out there suffer from that dreaded "hazy" look—washed out, imprecise, and lacking impact. I built this v2 merge because I wanted the absolute sharpest, consistent, most precise Illustrious model possible.
By carefully selecting and blending the highest-quality recent checkpoints (like Prefect Illustrious XL and ChosenMixXL) and meticulously tuning the merge ratios, I made sure to cook this properly. If you are looking for crisp, vibrant, and highly detailed results without the "fog", you are in the right place or I hope so lol.
⚙️ Recommended Generation Settings
For the best results, perfect saturation, and optimal sharpening, I highly recommend using the following parameters as your baseline.
CFG Scale: 5 (The sweet spot for balancing prompt adherence with perfect saturation and sharpness)
Steps: 20-40 (Ensures high detail and full convergence)
Sampler: DPM++ 2M SDE Heun (GPU Seeded)
Scheduler: KL Optimal (Nvidia AYS)
🚫 Negative Prompt Strategies
Golden Rule: Less is often more. Too many negative tags can confuse the AI and actually degrade image quality. I recommend starting minimal and only adding more if specific issues arise.
Choose the tier that best fits your workflow:
🟢 Minimalist (Highly Recommended)
Keeps the AI focused and prevents over-correction. Use this for the most natural, unhindered generations.
worst quality, bad quality
or
worst quality, low quality, lowres, bad, error
🟡 Standard (Balanced)
A great middle ground if you are seeing minor artifacting or stylistic issues without bogging down the prompt.
worst quality, low quality, lowres, blurry, pixelated, jpeg artifacts, oversaturated, watermark, signature, unfinished, sketch
🟠 Anatomy Focus
Use this variant if you are experiencing specific issues with character anatomy, hands, or eyes.
worst quality, low quality, lowres, blurry, pixelated, jpeg artifacts, mosaic blur, watermark, signature, unfinished, sketch, bad hands, distorted eyes, bad fingers
🔴 Aggressive (Use with Caution)
The "kitchen sink" approach. Use this only if the model is being particularly stubborn about errors, drafts, or unwanted elements.
worst quality, low quality, lowres, jpeg artifacts, cutoff, grainy, distorted, watermark, error, unfinished, sketch, draft, deformed, poorly drawn face, poorly drawn eyes, poorly drawn hands, fused fingers, bad hands, deformed hands, bad fingers, distorted eyes, censored, mosaic censoring, futanari, monochrome
(Note: You can remove "futanari" if that is a desired output).
Description
1. amanatsullustrious_v2.safetensors
2. hassakuXLIllustrious_v32.safetensors
3. prefectIllustriousXL_40.safetensors
4. oneObsession_v18.safetensors
Merges performed:
- Merged amanatsullustrious_v2.safetensors + hassakuXLIllustrious_v32.safetensors at 85/85.
- Merged oneObsession_v18.safetensors + prefectIllustriousXL_40.safetensors at 85/85.
- Merged the two intermediate results at 85/85, favoring the amanat/hassaku branch (so the final blend leans toward amanat + hassaku v32).
LoRAs:
- Applied StyleArtists\Enhance\Z_S_P_748cm... at 15% strength (model and CLIP).
- Applied StyleArtists\S_P_Momofuki_Rio_Illustrious... at 20% strength (model and CLIP).
Final blend and output:
- Merged the base blend with the LoRA outputs at 50/50 (final merge ratio 0.50).
- Used purecolorVAE_v10.safetensors as the VAE for finalizing colors.
- Saved checkpoint prefix: checkpoints/ComfyUI.
Notes:
- Goal was improved consistency on clothes; final model relies more on amanatsullustrious_v2 and hassakuXLIllustrious_v32.
- I avoided bringing in the lineart-heavy behavior from hassakuXLIllustrious_v13StyleB (not included in this merge).
- Did a lot of tests/finetunes and multiple saves during development, so some recorded ratios/labels might be slightly imprecise.