Attempt to recreate MadDogAi artstyle.

Recommended Settings:

• Strength: 0.8-1.3

• CFG: 4-5 (Non-Turbo recommended for best results)

• Steps: 24-30

• Upscaling: Recommended (Denoise 0.20-0.40)

Version Notes:

• Illustrious: Tends to add black lipstick due to the training set. May want to add lip color to prevent this. Trained on old images.

• Anima (v1): Fixed illustrious errors, trained on recent dataset.

• Anima (v2): Trained on newer images. Refined training settings and prompting for training. Works better than v1 on Anima Base 1, removing some of the "squished" feeling it had.

• Important Style Notes:

• Material Rendering: Black clothing will naturally render with a glossy, latex-like texture.

• Physique: The LoRA is heavily biased toward muscular, thick-thighed female characters.

• Lighting: This LoRA responds strongly to lighting prompts. If you want flat, bright lighting, you must explicitly prompt for it (e.g., flat lighting, very bright, daylight).

Negative Prompt Suggestions:

• To force pale skin: warm lighting, orange tint, red shadows, tan skin:1.2

• To prevent melted hands/limbs: bad anatomy, melted fingers, limb bleed

• To reduce latex look on non-black clothes: latex, shiny fabric, glossy,

• A lot of the images had sharp teeth, but it can generate normal teeth easily as long as sharp teeth isn't positively prompted. However, it is better to add sharp teeth in the negatives during upscaling since it can add them at that point.